BUDAPESTI MŰSZAKI ÉS GAZDASÁGTUDOMÁNYI EGYETEM

GÉPÉSZMÉRNÖKI KAR

Szerző neve: Kapás Nimród

Értekezés címe: Wind Effects on Natural Draught Dry Cooling Towers

Témavezető neve: Dr. Jászay Tamás

Értekezés benyújtásának helye (Tanszék, Intézet): Energetikai Gépek és Rendszerek

Tanszék

Dátum: 2005. február 25.

Bírálók: Javaslat:

Nyilvános vitára igen/nem

1. bíráló neve:

Nyilvános vitára igen/nem

2. bíráló neve:

Nyilvános vitára igen/nem

3. bíráló neve (ha van):

A bíráló bizottság javaslata:

Dátum:

(név, aláírás)

a bíráló bizottság elnöke

NYILATKOZAT

Alulírott Kapás Nimród kijelentem, hogy ezt a doktori értekezést magam készítettem, és

abban csak a megadott forrásokat használtam fel. Minden olyan részt, amelyet szó

szerint, vagy azonos tartalomban, de átfogalmazva más forrásból átvettem, egyértelműen,

a forrás megadásával megjelöltem.

Az értekezés bírálatai és a védésről készült jegyzőkönyv a későbbiekben a

Gépészmérnöki Kar dékáni hivatalában elérhetőek lesznek.

Budapest, 2005. február 25.

STATEMENT

I, Nimród Kapás undersigned, declare that I have prepared this Ph.D. dissertation by

myself, and I used only the specified sources in it. Every part, which I included here from

other sources word for word, or with the same content but rephrased, I marked it by

indicating the source unequivocally.

The critiques of the dissertation and the report of the defence will be available in the

Dean’s Office of the Faculty of Mechanical Engineering.

Budapest, 25 February 2005.

Kapás Nimród

Budapest University of Technology and Economics1782

Department ofEnergy Engineering

Wind Effects on Natural Draught Dry Cooling Towers

Ph.D. dissertation

Supervisor: Author:Tamás Jászay Nimród Kapás

Dr. Techn. Habil. Ph.D. Candidate for theProfessor emeritus Ph.D. degree

BUDAPEST, 2005

CONTENTS

i

CONTENTS

NOMENCLATURE iv

PREFACE viii

1. INTRODUCTION 1

1.1. Electricity generation..........................................................................1

1.2. Thermal power plants in general........................................................3

1.3. Power plant cooling systems...............................................................4

1.3.1. Indirect dry cooling system..................................................................... 5

1.3.2. Natural draught wet cooling towers ..................................................... 13

1.4. Main objectives and relevance of the present Ph.D. work ...............15

2. LITERATURE OVERVIEW 17

2.1. Full-scale experience ........................................................................17

2.2. Scale model laboratory tests.............................................................21

2.3. Numerical simulations ......................................................................31

3. ON-SITE MEASUREMENT OF WIND EFFECT 43

3.1. Main data of the measured cooling system.......................................43

3.2. Execution of the measurement ..........................................................44

3.2.1. Calibration of the measuring instruments ............................................ 44

3.2.2. Operation of the measuring system ...................................................... 45

3.3. Evaluation of the thermal performance of the cooling system .........49

3.3.1. Another evaluation method................................................................... 51

CONTENTS

ii

3.4. Wind effect curve obtained from full-scale measurements ...............51

4. COMPUTATIONAL FLUID DYNAMICS MODELLING 55

4.1. Geometry...........................................................................................55

4.2. Mesh ..................................................................................................57

4.3. Model settings ...................................................................................59

4.3.1. Grid manipulation................................................................................. 59

4.3.2. Turbulence modelling ........................................................................... 59

4.3.3. Material properties ............................................................................... 59

4.3.4. Operating conditions ............................................................................ 59

4.3.5. Wall boundary conditions..................................................................... 60

4.3.6. Thermal modelling of heat exchangers................................................. 60

4.3.7. Pressure drop in heat exchanger zones ................................................ 63

4.3.8. Turbulence modelling in the heat exchanger zones.............................. 63

4.3.9. Wind profile .......................................................................................... 65

4.3.10. Fan boundary condition........................................................................ 66

4.3.11. Air tunnel of peak coolers..................................................................... 66

4.3.12. Tower X-legs ......................................................................................... 66

4.3.13. Louvers.................................................................................................. 67

4.3.14. Solver settings ....................................................................................... 67

4.4. Solution initialisation and monitoring..............................................67

4.5. Results ...............................................................................................69

4.5.1. Flow visualisation................................................................................. 69

4.5.2. Wind effect curve .................................................................................. 75

4.5.3. Thermal performance ........................................................................... 77

4.5.4. Dimensionless numbers of flow similarity ............................................ 79

4.6. Grid adaption....................................................................................80

4.7. Data about the computer hardware..................................................81

CONTENTS

iii

5. WIND EFFECT IMPROVEMENT POSSIBILITIES 82

5.1. Measures on the water side...............................................................82

5.1.1. Increasing the water flow rate to the tower.......................................... 82

5.1.2. Controlling the water flow into different heat exchanger sectors ........ 84

5.2. Measures on the air side ...................................................................87

5.2.1. Measures at the air inlet opening ......................................................... 87

5.2.1.1. CONTROLLING THE LOUVERS........................................................................875.2.1.2. BAFFLE PLATES ............................................................................................88

5.2.2. Measures at the tower outlet................................................................. 95

5.3. Economical evaluation......................................................................97

6. SUMMARY OF NEW RESULTS 98

6.1. Thesis 1. ............................................................................................99

6.2. Thesis 2. ............................................................................................99

6.3. Thesis 3. ............................................................................................99

6.4. Thesis 4. ............................................................................................99

6.5. Thesis 5. .......................................................................................... 100

REFERENCES 101

List of Own Publications 110

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT 111

ÖSSZEFOGLALÓ MAGYARUL 117

SUMMARY IN ENGLISH 118

ABSTRACT 119

NOMENCLATURE

iv

NOMENCLATURE

Symbols

A [ m2 ] – area

Ar [ - ] – Archimedes number

Cpi [ - ] – dimensionless tower inlet stagnation pressure loss coefficient

cp [ J/(kg·K) ] – specific heat at constant pressure

cw [ J/(kg·K) ] – specific heat of water

Cw [ - ] – wind effect coefficient

D [ m ] – diameter

g [ m/s2 ] – gravitational acceleration

H [ m ] – cooling tower height

hi [ m ] – tower intake height

k [ J/kg ] – turbulent kinetic energy

ma [ kg/s ] – air mass flow rate

mw [ kg/s ] – cooling water mass flow rate

p [ Pa ] – pressure

pb [ Pa ] – barometric pressure

q [ W/m2 ] – heat flux

Q [ MW ] – dissipated heat

Q* [ MW ] – corrected heat dissipation

r [ m ] – radius, radial coordinate axis

Ta [ ºC ] – air temperature

Ta1 [ ºC ] – cooling tower inlet air temperature

Ta2 [ ºC ] – cooling tower outlet air temperature

Tcell [ ºC ] – cell temperature

TH2 [ ºC ] – fluid temperature exiting the radiator

THX [ ºC ] – radiator temperature

Tin [ ºC ] – inlet temperature of the coolant in a macro

NOMENCLATURE

v

Tin0 [ ºC ] – inlet temperature of the coolant in the first macro

Tout [ ºC ] – outlet temperature of the coolant in a macro

Ts [ ºC ] – saturated steam temperature

Tw [ ºC ] – water temperature

Tw1 [ ºC ] – warmed water temperature

Tw2 [ ºC ] – cooled water temperature

v [ m/s ] – velocity

vi [ m/s ] – inlet air velocity to the cooling tower

v10 [ m/s ] – wind speed measured at 10 m above ground level

v11 [ m/s ] – wind speed measured at 11 m above ground level

v96 [ m/s ] – wind speed measured at 96 m above ground level

x [ m ] – coordinate axis

z [ m ] – coordinate axis (in the vertical direction)

α [ W/(m2·K) ] – heat transfer coefficient

Δ [ - ] – difference

ΔQa [ MW ] – correction for cooling tower heat rejection due to the deviation ofambient air temperature from the nominal condition

ΔQb [ MW ] – correction for cooling tower heat rejection due to the deviation ofbarometric pressure from the nominal condition

ε [ m2/s3 ] – turbulent dissipation rate

ε HE [ - ] – heat exchanger effectiveness

ϕ [ º ] – tangential coordinate axis

Φ [ s/m ] – density function of wind velocity

η [ % ] – effectiveness

ρ [ kg/m3 ] – density

ϑt [ ºC ] – Initial Temperature Difference of the tower

θ [ º ] – central angle of the tower (θ = 0 º points in the -x coordinatedirection)

Subscripts

0 – no-wind conditions

NOMENCLATURE

vi

m – measured

n – nominal

t – tower

w – water, wind

∞ – undisturbed flow

Abbreviations

2D – two-dimensional

3D – three-dimensional

CCPP – Combined Cycle Power Plant

CFD – Computational Fluid Dynamics

Co. Ltd. – Company Limited

CPC [ MW/K ] – Cooling Plant Capacity

CPU – Central Processing Unit

EdF – Electricité de France

EGI – an abbreviation remained from the former name of my employer:Institute for Energetics (in Hungarian: EnergiagazdálkodásiIntézet)

FVM – Finite Volume Method

GT – Gas Turbine

HRSG – Heat Recovery Steam Generator

ITD [ ºC ] – overall Initial Temperature Difference

ITD0 [ ºC ] – nominal overall Initial Temperature Difference (in no-windconditions)

MBV – Modified Bouver-Vogel

PC – Personal Computer

PGT – Performance Guarantee Test

Ph.D. – Doctor of Philosophy (Doctorate degree)

PS – Power Station

RNG – Renormalisation Group

ST – Steam Turbine

NOMENCLATURE

vii

TPP – Thermal Power Plant

TTD [ ºC ] – Terminal Temperature Difference, sub-cooling in the condenser

USSR – Union of Soviet Socialist Republics

PREFACE

viii

PREFACE

After graduating with honours in 2000 at the Budapest University of Technology andEconomics, Faculty of Mechanical Engineering, I started to work at the EGI -Contracting Engineering Co. Ltd. as a design engineer. In the course of designing thecomponents of new power plant cooling systems, preparing the related sizing calculationsand surveying the manufacturing, assembling and testing activities at EGI’s productionplant in Jászberény, I obtained detailed working knowledge in this field.

At the same time, I continued my postgraduate studies as a correspondence student atthe organised three-year Ph.D. course of the Faculty of Mechanical Engineering, withconsultation support from the Department of Energy Engineering.

In the spring of 2001, I have begun to deal with the wind effect on natural draught drycooling towers, firstly by reviewing the earlier experiences of EGI which relied mostly onfull-scale measurements and scale model experiments in the wind tunnel in Jászberény.Later, I flew to Turkey with one of my colleagues to perform new on-site measurementson an operating dry cooling tower in June-July, 2001. After that, I was shifted to theResearch and Development Section at EGI, so I latched on to the actual research worksand assisted in the development of an advanced computer program for the evaluation ofthese measurements.

In the meantime, I studied the economic and environmental aspects of different energytechnologies, too, and contributed to the Student Programme of the 18th World EnergyCongress & Exhibition organised by the World Energy Council in Buenos Aires on 21-25October, 2001.

In 2001 EGI won government subsidy in the framework of National Research andDevelopment Programs 2001, called the Széchenyi-Plan, by which several aspects of thecooling tower design were refined in consortium collaboration with other companies anddepartments from the Budapest University of Technology and Economics.

From 2002 January, EGI purchased the license for a Computational Fluid Dynamics(CFD) software called FLUENT. This allowed inter alia numerical calculations of thecooling tower in wind to be performed. The results were compared with the on-sitemeasurements, and the correspondence was more than adequate. Thereafter, I madeseveral simulations in order to augment our knowledge about the behaviour of naturaldraught cooling towers in wind. The results are summarised in this Ph.D. work.

Here I would like to express my thanks to my supervisor professor Tamás Jászay fromthe Department of Energy Engineering, Budapest University of Technology andEconomics, for his valuable ideas, comments, continuous encouragement and help toobtain subsidies for covering the expenses incurred in connection with my doctorate.

PREFACE

ix

I also render thanks to Mr. László Ludvig, Head of Design and EngineeringDepartment of the Heller Power and Process Cooling Systems Business Line at EGI -Contracting Engineering Co. Ltd., by whom I was allowed, beyond I performed myeveryday engineering tasks, to study the effect of wind on cooling towers at theworkplace and to use the technical background of EGI. I am grateful to Mr. Gábor Csaba,Head of Research and Development Section at EGI, for his help and expertise provided inthe evaluation of full-scale measurements.

I acknowledge the consultancy support of the educational staff of the Department ofFluid Mechanics, Budapest University of Technology and Economics, relating tomodelling fluid flows in FLUENT.

Last but not least thanks to my parents and friends, to whom I was missing duringPh.D. studies, and they tolerated this with patience.

INTRODUCTION

1

1. INTRODUCTION

This chapter will demonstrate at first the role of cooling systems in electricity generatingpower plants, after that the main components and operating principles of the Heller® typeindirect dry cooling system to be dealt with in this Ph.D. study will be introduced.Finally, the objectives of my analyses are presented.

1.1. Electricity generation

Within energy in general, electricity plays a special, indispensable and steadily increasingrole in our life. The industrial, residential, transport, medical and telecommunicationsectors would meet with heavy losses if power-cut occurred even just for a few minutes.Therefore, reliable energy supply is needed for the mankind to maintain a sustainablegrowth globally.

Electricity is generated from different kinds of primary energy sources, which arelisted below:

Fossil fuels (coal, oil, natural gas),

Nuclear,

Renewable energy (solar, hydro, wind, geothermal, ocean wave and tidal),

Other (combustible renewables, biomass and waste, i.e. wood, agricultural by-products, animal refuse and wastes, gas and liquids from biomass, sulphite lyesfrom the manufacture of paper, municipal, industrial and hospital waste).

Figure 1-1. Evolution of world electricity generation* by fuel, TWh* Excludes pumped storage.

** Other includes geothermal, solar, wind, combustible renewables and waste.Source: http://www.iea.org/statist/keyworld/keystats.htm

15000

12500

10000

7500

5000

2500

01971 1974 1977 1980 1983 1986 1989 1992 1995 1998

Thermal Nuclear Hydro Other**

INTRODUCTION

2

From the beginning of commercial energy supply to different end-users, fossil fuelthermal power stations have done the bulk of electric power generation and this cannot beexpected to change during the next decades (see Figure 1-1.). Among fossil fuels, coalhas the largest geological reserves with the most evenly distributed occurrence in theworld. However, the use of coal involves environmental concerns – emission ofgreenhouse gases, acid rain – however, recently many types of clean technologies(Fluidised Bed Combustion, Integrated Gasification Combined Cycle, Flue GasDesulphurisation, sorbent injection, spray drying, reburning, combined SOx/NOx, lowNOx combustion, post-combustion NOx, Selective Catalytic Reduction, CO2 scrubbers,physical coal cleaning, CO2 sequestration) help to adapt it a lot better to environmentalrequirements.

Although larger oil-fields are located only in certain regions on the Earth andalternative drives for vehicles are already in the research & development phase, due to itshigh energy density (MJ/litre) and easy tractability, oil is used as a predominant motorfuel in the transportation sector throughout the world. The heavy fuel oil residuals(distillation residues from the oil refineries) and other oil products can be used in oil-firedthermal power plants. Oil has also environmental problems (air and sea pollution).

Natural-gas-fired Combined Cycle Power Plants (CCPP) are one of the most advancedtechnologies in electricity generation. The enthalpy of the gas turbine exhaust gases (withcc. 600-800 ºC) is utilised in a Heat Recovery Steam Generator (HRSG), which producessteam for a steam turbine without burning of additional fuel in the steam cycle. Thethermal efficiency of these power plants is cc. 0.58 (for comparison, the thermalefficiency of a spark ignition Otto-motor, a compression ignition Diesel motor and anuclear power plant is 0.3-0.35, 0.35-0.45 and 0.33, respectively), furthermore, lesspollutants and heat load per generated electricity are emitted in the environment.

Nuclear energy has cheap and potentially large fuel resources (reserves can be largelyextended by breeder implementation) and the nuclear fission power plants emit negligibleamounts of greenhouse gases during operation. However, due to reactor safety aspectsthis technology faces public acceptability problems and the fears of the risk of the finaldisposal of radioactive waste and technology proliferation often lead to keep off investingin the nuclear energy.

Renewables are „zero-emission” technologies, however, their economics dependsstrongly on site conditions and also have their own environmental problems. Largehydroelectricity is a proven technology, capable of generating thousands of MWs.Nevertheless, a part of public opinion has turned against hydropower because of the damsthey require and in industrialised countries the potential sites have mostly been exploitedalready.

The output of the wind turbines is proportional to the wind speed (intermittentavailability). Therefore, wind energy is suitable only to coastal areas and high ridges,furthermore, the individual machines are unsightly, noisy and take up a lot of valuable

INTRODUCTION

3

land. Their intermittent electricity production needs keeping full capacity reserve intraditional power stations.

Photovoltaic cells can convert diffused light and direct sunlight both into electricity,but this technology is more productive and cheaper just at lower latitudes. The costs ofthe solar electricity are high, but prices started to decrease. However, solar energy requirelarge surface area and only high capacity plants would be viable.

The utilisation of biomass energy is also a proven technology and the sources areplentiful in waste products. Furthermore, this technology offers some independence fromfossil fuels and bio-forests can act as sinks for carbon-dioxide. Biogas can be burned incombined cycle plants (co-firing with fossil fuels). The disadvantages are that the burningof biomass produces gaseous and liquid waste and the collection, transportation andstorage of biomass is expensive.

1.2. Thermal power plants in general

In thermal power plants the heat input can be obtained by means of fuel combustion(fossil fuels and biomass), nuclear fission, solar radiation (solar thermal) andunderground high pressure steam or warm water and hot dry rocks (geothermal plants).The thermal energy is firstly converted into mechanical work by gas or steam turbines,which is then transformed into electricity by electric generators.

In gas turbine cycles, air is brought to high pressure by a compressor, then the heatinput eventuates and the high pressure and high temperature air enters the turbine. Whilethe air is expanding and flowing through the turbine blading, the shaft of the gas turbine(which is coupled together with the shaft of the compressor) is driven. Usually, theexhaust gases are discharged into the atmosphere (open cycle gas turbines). In thistechnology there is no need for re-cooling the working fluid because the cycle is open.

Similarly, in steam turbine cycles water is brought to high pressure by a feedwaterpump, then high temperature overheated steam is generated in a boiler by means of theheat input. After that steam is flowing through the turbine and is expanding at theexpense of its internal energy, while its specific volume is increasing and the surfaces ofthe turbine blading are rotated around, so mechanical work is produced on the steamturbine shaft.

Modern high efficiency steam cycles are closed, i.e. after the turbine the steam flowsinto a condenser, where vacuum is maintained. In this way, the enthalpy and pressuredifference of the steam utilised in the turbine is higher than it would be if dead steam wasexhausted into the atmosphere at about 1 bar. In the condenser, the steam undergoes aphase change (condensation), so the water is recovered and flows on to the feedwaterpump. However, this cycle needs a significant amount of low-level heat to be rejected (asaccording to the Second Law of thermodynamics, thermal energy cannot be convertedcompletely into mechanical work) to the atmosphere or natural water bodies in order to

INTRODUCTION

4

allow the condensation to take place. The temperature of the condensation should be lowfor the better thermal efficiency of the plant, but the temperature of the condensing steamin the condenser must be somewhat higher than the temperature of the cooling air or thecooling water.

Alternatively, the latent (condensation) heat of the turbine exhaust steam can beutilised for heating purposes (Combined Heat and Power plants, co-generation plants)instead of rejecting it into the environment. While industrial heat demand is steady ingeneral, demand for space heating is by its nature seasonal, therefore, an auxiliarycooling system is needed to keep the power station running also in space heating off-seasons.

Summing up: condensing steam power stations generate the lion’s share of electricity.To carry out condensation (heat rejection) they need cooling. The temperature level ofheat rejection has to be as near to the ambient temperature as possible. The amount ofheat to be rejected is very big. All this translates to the fact that the cooling systemrepresents a decisive component of a steam power station.

There are many types of power plant cooling systems, and their application depends onthe yield of available natural water sources nearby the power plant. Therefore,availability of adequate amount of cooling water is an important factor in choosing theoptimal site for a thermal power station, but the transport distances and costs of the fuelas well as transmission distances and costs of generated electricity have to be alsoconsidered.

1.3. Power plant cooling systems

The main types of the cooling systems are:

a) Fresh-water (once-through) cooling is applied if the power plant is located in thevicinity of a natural source of water (sea or river). This is the most advantageousoption in respect of the thermal efficiency of the plant. In this case the coolingwater is taken from the environment, and after the cooling process, water isdischarged back into the natural bodies of water. To avoid the direct mixing of thehigh purity steam condensate (of boiler feed water quality) with the much lowerquality cooling water, (containing also pollutants and dissolved gases which wouldincrease the condenser pressure), surface condenser must be used, in which thecooling water flows in tube bundles and the steam flows and condensates outsidethe tubes.

b) Evaporative cooling requires less natural cooling water than once-through cooling.After warming up in the surface type condenser, the cooling water flows into anartificial or natural cooling pond with or without spraying equipment (likewaterworks) or in a natural or mechanical draught cooling tower. The cooling watergets in direct contact with ambient air on the surface of the pond or on the surfaces

INTRODUCTION

5

of the packing of the wet cooling tower, and is re-cooled in a simultaneous heat andmass transfer process by convective and latent (evaporative) heat transfer. The re-cooled water is returned into the condenser by pumps. Water losses are only due toevaporation, carry over of drops by wind and blow down. It can be said that thesupplemental water demand is nearly equal to the steam flow to the condenser,because the amount of the condensed steam is nearly equal to the evaporatedcooling water. Depending on the size of the contact surface, water temperatureapproaches the wet bulb temperature of the ambient air.

c) Dry cooling systems exclude the direct contact of the cooling water and the air(cooling water or steam flows in finned tube bundles and ambient air flows throughthe fins), so water losses mentioned at evaporative cooling are avoided. As there isno need for significant make-up water supply, the power plant can be located at fuelsource (mine or oil refinery). The environmental impact is lower than in case ofevaporative cooling (no vapour-plume emission originating from the evaporatedcooling water), and so, less public resistance can be expected against theconstruction. However, the cooling process is less efficient than with wet cooling,because the cooling water temperature can reach only the dry bulb temperature ofambient air in ideal case. Dry cooling requires 3-4 times more cooling air flow thanwet cooling for same steam turbine performance. Generally, the exhaust annulusarea of the low pressure turbine must be smaller with a corresponding higher endloading at dry cooled power plants, than at wet cooled ones.

There are two types of dry cooling systems:

– indirect dry cooling „Heller® System”,

– direct dry cooling (Air Cooled Condenser).

d) Hybrid and combined cooling systems combine the advantages of different coolingsystems (combined once-though/wet, hybrid wet/dry, combined dry/wet).

1.3.1. Indirect dry cooling system

A schematic diagram of the Heller® type indirect dry cooling system (named after the lateprofessor László Heller, who invented the system in Hungary in the late 1940’s) is shownin Figure 1-2.

As cooling water is circulating in a closed circuit, direct contact jet condenser can beused. The exchange surface is formed by water jets sprinkled out from spray nozzles(water jets leaving the holes of the nozzles impact to baffle plates thus making waterfilms), so the steam is mixing directly with cooling water which is of boiler feed waterquality. In the last phase of condensation about 5 % of the steam flow gets to the after-cooler section where steam, along with its air/non-condensable content is flowing throughwater curtains trickling down from a perforated, cascade tray system in counter flow

INTRODUCTION

6

direction, and further condensation goes on1. At the end of the after-cooler the airevacuation system removes from the condenser the air and other non-condensable gases(see Figure 1-3.). The evacuation system usually comprises a steam jet air ejector withsilencer.

HRSG

Steam turbine

Generator

Condensate booster pump

Direct contact jet condenser

Cooling water circulating pump

Motor Recovery hydro turbine

Louvers

Heat exchangers

Dry cooling tower

Air flow

Ts Tw2

Tw2

Tw1

Tw1

Ta1 Ta2

Figure 1-2. Schematic drawing of the indirect dry cooling Heller® system in a CCPP

Figure 1-3. Operation of the direct contact jet condenser

1 A more detailed description and analysis of direct contact condensers are available inter alia inÁ. BAKAY, T. JÁSZAY: High Performance Jet Condensers for Steam Turbines. Transactions of theInternational Heat Transfer Conference, Toronto, 1978.

SteamDivision plate

Water header

Verticalwater films

Air extraction

After-coolersection

Perforated trays

Nozzles

INTRODUCTION

7

The direct contact jet condenser is a highly efficient equipment – there is only amarginal sub-cooling (difference between the saturated steam temperature, Ts, andwarmed water temperature, Tw1, called as Terminal Temperature Difference, isadvantageously near zero or at least less than 0.3 °C for direct contact jet condensers andbetween 1.8-4 °C for surface condensers), and much less material is required forconstruction, as there are no heat exchanger tubes. Furthermore, because of its simplerconstruction, the jet condenser’s reliability is much higher and its maintenance cost lowerthan those of the surface condensers. The direct contact jet condenser is more efficient,smaller, lighter and cheaper than a surface condenser with its tube bundles.

The condensed steam and the warmed water are collected in the bottom „hotwell” partof the jet condenser. Cc. 2 % of this flow - corresponding to the amount of steam condensed– is fed to the boiler feed water system by simple booster pumps. The major part of the flowis delivered to the cooling tower by circulating pumps for cooling. During operation, thecooling duty is performed by the heat exchangers, where cooling air flow is induced by anatural draught cooling tower: while flowing through the heat exchangers, the temperatureof the air increases, consequently, the air inside the tower with smaller density than that ofthe ambient air outside the tower will rise upwards due to buoyancy forces.

Natural draught cooling towers may be constructed from welded/bolted steel structurewith aluminium cladding, but reinforced concrete hyperbolic shells are also applicable (seeFigure 1-4. and 1-5.). Steel towers are preferred to concrete shells at sites of high seismicity.The painted steel structure also safely resists corrosion (the air inside the cooling tower isnormally approx. 20 ºC hotter, and hence, drier than ambient air).

Figure 1-4. Steel structured cooling towers Figure 1-5. Reinforced concrete cooling towersShahid Montazeri TPP 4 × 210 MWe (Esfahan, Iran) 1200 MWe Trakya CCPP (Turkey)

Heat exchangers can be arranged either vertically around the perimeter of the coolingtower or horizontally in the inlet cross-section (see Figure 1-6. and 1-7.). In the Heller®

system, the vertical arrangement is used mainly.

INTRODUCTION

8

Figure 1-6. Heller type natural draught cooling tower

Source: VAN STADEN ET AL. (1996)

Figure 1-7. Cooling tower with horizontally arranged heat exchangers

Source: VAN STADEN ET AL. (1996)

Usually two hydraulic machine groups are connected in parallel in the cooling watercirculating system, each consisting of a cooling water circulation and extraction pump, arecuperating hydraulic turbine and a driving electric motor - all three machines beingmounted on a common shaft connected together by plug type coupling. Thus powerrecovered by the hydraulic turbine (utilising the relatively high pressure of cooling waterarriving from the top of heat exchangers) provides part of the pumping power, the remainderbeing supplied by the motor. If one of the two pumping units is off, the same heat rejection

INTRODUCTION

9

at approx. 5.5 ºC condenser temperature increase, or 80 % heat rejection with the originalcondenser temperature still can be maintained with the pump remaining in operation. As thecooling water pumps get the cooling water from the condenser where vacuum is prevailing,low speed (~ 425 min-1) pumps are used in a pit at elevation lower than that of the condenserin order to avoid cavitation.

The pumps extract water from the direct contact jet condenser and raise water pressure tosuch a value that under all steady state conditions, even at the highest point of the coolingdeltas, it will exceed atmospheric pressures. This ensures that if a leak should occur, air willnot be drawn into the system, with the consequent reduction in vacuum. On the other hand,due to the overpressure the location of the leak will be immediately apparent. The requiredwater level above the top of heat exchangers is maintained and controlled by the hydraulicturbines, and in the few meters long piping from the hydro turbines to the condenser thepressure is again low.

The cooling water is flowing in an underground distributing ring-piping in the tower,from which the sector inlet pipelines are branching towards the groups (or sectors) of heatexchangers, where the cooling columns are connected in parallel at the water side.Similarly, the cooled water after the heat exchangers flows back through the sector outletpipelines to the underground collecting ring-piping in the tower (see Figure 1-8.). If e.g.one of six parallel cooling delta sectors is disconnected, approximately 90 % heatrejection with unchanged condenser temperature or 100 % heat rejection with 2.5 ºCcondenser temperature increase can be maintained.

Figure 1-8. Piping system inside the cooling tower

The self-supporting erection unit of the heat exchanger bundles is called cooling delta. Ithas a rigid framework of triangular cross-section, two sides of which are formed from 10, 15

Tower gate Underground ring-piping

Sector inlet/outlet lines

Tower X-legs

Heat exchangers

Storage tanks

INTRODUCTION

10

or 20 m high, 2.4 m wide heat exchanger panels (columns) with an angle of approx. 58° (thecooling delta angle), while the third side is left open for air inlet usually equipped withelectrically actuated louvers for air flow control. Connection between the distribution pipingand the coolers is made by flexible rubber hoses.

The air is flowing into the cooling deltas across the louver field, and is passing throughthe cooling columns dissevering in two directions (see Figure 1-9.). The warm waterenters the cooling columns through the bottom header (in which a bulkhead is placedwith two separated passes), flows up in the cooler tubes, then turns down in the topheader and flows down in the cooler tubes. So while it is re-cooled, the cooling watergoes to the cold part of the bottom header in a cross-counter flow relative to the air flow.From the bottom header, the cooling water flows back into the piping system through thecold water outlet stub.

Cooling delta

Air flow

Louver

Cooling column

Air flow

Steel structure

Louver

Air flow

Steel structure

Cooling column

Cooler bundle

Plate fin

Cooler tube

Bottom header

Warm water inlet

Cold water outlet

Figure 1-9. Cooling delta – Forgó type water-to-air heat exchangers

In the Forgó type heat exchangers tubes are arranged in staggered pitch forming a jointcooler matrix with perforated fin plates (see Figure 1-10.). Assembling of the all aluminiumheat exchanger surface is an entirely cold process (no welding or soldering is used) – a steelball is pushed through the cooler tubes, so the tubes dilate and get nipped into the collaredholes of the plate fins. Consequently, there is no danger of corrosion due to the remainingwelding or soldering fluxes. Due to the mono-metal design the bi-metallic corrosion is alsoavoided.

The heat exchanger surface undergoes also the MBV (Modified Bouver-Vogel)treatment, which results chemically in a thicker protective oxide layer. Based on the

INTRODUCTION

11

experiences, the life of air coolers can be expected to be over 30 years with proper cleaning(by pressurised water) and maintenance.

Figure 1-10. Structure of the heat exchanger surface

The Heller system can be optionally equipped with specially configured, additional coolersurfaces suitable for dual-purpose operation: wintertime preheating of cooling deltas by hotair-blast during cold start-up and summertime cooling capacity enhancement with amoderate amount of deluge water evaporation on the finned surface of cooler bundles. Thisenhancement generally results in a condenser temperature drop of about 2.5-3 ºC, addingabout 1 % to the plant net generation efficiency.

The pre-heaters/peak coolers are induced mechanical draught cooler cells connected inparallel with cooling deltas and are divided into separate sectors situated adjacent to themain cooling deltas inside the natural draught cooling tower. Pre-heater/peak cooler groupsnormally take part in the cooling operation by means of the tower natural draught. Fans arerunning only during deluged operation in the hottest summer peak periods and duringpreheating in winter, in the latter case running in inverse direction with the louvers of themain cooling deltas fully closed (see Figure 1-11.).

Figure 1-11. Operating principle of pre-heater/peak cooler cells

INTRODUCTION

12

The fact that it is worth investing research and development work in the indirect drycooling system is supported by the advantages of the Heller system:

Flexible site arrangement (unlike at direct Air Cooled Condensers, where the heatexchangers must be located as near as possible to the steam turbine in order toavoid additional pressure losses in the steam flow and air infiltration into the longermain steam duct under vacuum, at indirect dry cooling locating air coolers adistance away means elongating cooling water piping, which causes only smallincrease in costs – piping is inexpensive and pressure loss in straight pipelines isinsignificant).

Safety against freezing danger (easy control of cooling air flow rate by the louvers;emergency draining of the heat exchangers by the control system; pre-heater cells –absolute safe start-up and shut-down at extremely cold climates).

Cooling system start-up with zero steam flow – cooling water by-pass circulation.

Water saving (unlike wet towers, dry cooling systems require no significantamounts of make-up water; a 200 MWe steam turbine unit equipped with wetcooling tower has water consumption equivalent to the continuous water demand ofa medium size city of 100,000 inhabitants. Thus deciding for power plant drycooling may give chance for further development of a whole region leaving theavailable water resources to use in the residential, industrial and agriculturalsectors).

Environment friendly – easy permitting for construction. In contrast to dry cooling,wet towers discharge concentrated cooling water blow-down, which may pollutethe surroundings, and once-through cooling systems discharge warmed water backto natural bodies of water which may damage the aquatic flora and fauna because athigher temperatures the solvability of oxygen in water is decreased.

A total of 17,000 MWe power plants are in service and under construction with theHeller system including units operating under extreme ambient conditions (as coldas -62 °C or some at +50 °C).

High availability and reliability: clean cooling water cycle precludes condenserfouling leading to the deterioration of heat rejection capacity and condenserclogging (by algae or mussels/sea weed) that plagues wet cooled and once-throughseawater cooled units with increased unavailability.

Large cooling water fill (approximately 3,000 m3) helps smooth out fluctuations ofcooling water temperature variations due to wind-gusts and boiler make-up waterpurity when demineralisation plant occasionally malfunctions.

Natural draught:

INTRODUCTION

13

– Low auxiliary electric power consumption with natural draught (approx. 2 %improvement in steam cycle efficiency by the elimination of self consumptionfor cooling air moving by fans).

– Low maintenance requirements (due to absence of air moving equipment, theonly continuously moving components of the system are the cooling watercirculating pump units requiring brief maintenance during scheduled plantshut-down).

– Low noise pollution.

– No hot air recirculation (in contrast with mechanical draught coolers, wherethe air inlet and outlet openings are close to each other, warmed airdischarged from the tower may not re-enter the heat exchangers at the intakedue to wind, resulting in performance deterioration).

– Integrating boiler flue-gas stack and desulphurisation equipment inside thecooling tower sharply reduces stack costs (cooling tower draughtsuperposition results in lower flue-gas stack height) and ground levelconcentrations of pollutants emitted through the power plant stack (flue-gasmeets the large quantity of warm, dry cooling air inside the cooling tower, inwhich the flue-gas thins down and gets to high altitudes).

1.3.2. Natural draught wet cooling towers

In the engineering science literature, a number of studies analyse the effect of wind onnatural draught wet cooling towers. These surveys are very useful in understanding thewind effect on dry cooling towers, too. Therefore, a brief introduction of this type ofcooling towers is outlined below.

Figure 1-12. Outline drawing of a cooling system with counter-flow natural draught cooling tower

Water consumption:Cc. 1.8-2.5 % of the circulated water flow rateAuxiliary power consumption:Cc. 0.8 % of the steam turbine performance+ cc. 0.7 % if there are ventilators

GeneratorSteam turbine

Surface typecondenser

To boiler Coolingwaterpump

Motor

Coolingtower

Wet air(water-plume)

Drift eliminator

Water sprayers

Packing

Cold air

Cold water basin

Make-up water

Rain zone

INTRODUCTION

14

The main components of a wet tower are the water sprayers, inner-shell packing, drifteliminator and the water collection basin (see Figure 1-12., 1-13. and 1-14.). The warmwater is sprayed through a grid of sprinklers over the packing. The packing or fill is amulti-layered lattice with large specific air-to-water contact surface, which obstructs thefree fall of water extending by that the residence/heat and mass transfer time. Thepacking breaks up water flow into droplets, increases the contact area and contact timewith air, and therefore improves the heat transfer rate and efficiency of the cooling tower.While the cooling water is flowing down the packing, it is cooled by the air colder than it.Below the packing, the water is further cooled in the so called „rain zone” and is thencollected in the basin at the bottom of the cooling tower for returning into the condenserby pumps. The buoyancy forces and chimney effects induce natural draught in the tower:the air flowing sideways into the tower is warming up and rises due to its decreaseddensity and leaves the tower at its upmost cross-section.

Similarly to natural draught dry cooling towers, the heat (and mass) exchanger packingin wet towers can be arranged either horizontally in the inlet cross-section of the tower(called as counter-flow towers – the water and the air are flowing in counter flowdirection) or vertically around the circumference of the tower (called as cross-flow towersaccording to the water and air flow).

Wet cooling towers provide the „cooling” effect to the water mainly through theevaporation of some of the water during its direct contact with air. The cooling towerindustry typically quotes that the fraction of the heat removed from the water byevaporation is approximately three-quarters of the total heat rejected to the atmosphere(ASME (PTC) 23-12986 (R1992)). Interestingly, this fraction can exceed 100 % if theinlet water temperature is lower than the inlet dry bulb air temperature. In this case thesensible heat transfer will be negative, that is from the air to the cooling water, and underthese conditions both the air and the water are cooled. This phenomenon takes placeunder extreme hot and extreme dry ambient air parameters. Sensible heat transferinvolves an increase in the dry bulb temperature of the air, and is distinguished from thelatent heat transfer involving a change in the moisture content of the air, defined as themass of water vapour per unit mass of air.

In the literature, the effectiveness of the cooling towers is generally characterised bythe temperature approach, which is the difference between the cold water temperature ofthe tower and the ambient wet-bulb temperature (dry-bulb for dry cooling towers)corresponding usually to a constant heat load.

Wet cooling towers require smaller plot area than dry cooling systems, but the lifetimeof heat exchanger packing used in wet systems is lower (~ 10 years) and due toevaporation, visible plume is emitted at the top of the tower (see Figure 1-15.). The latterphenomenon may (and usually does) arise protest on behalf of the neighbourhoods.(While the same, protesting people enjoy the fleecy clouds on the sky.)

INTRODUCTION

15

Figure 1-13. Film type counter-flow fill Figure 1-14. Profile type drift eliminator

Figure 1-15. Lignite-fired Schwarze Pumpe PS with wet cooling towers, Germany, 2 × 800 MWe

1.4. Main objectives and relevance of the present Ph.D. work

A cooling system is an integrated part of a power plant with steam cycles. The thermalefficiency of the plant is higher if the difference between the highest and the lowesttemperatures in the cycle is greater. The highest applicable temperature is constrained bythe deterioration of material mechanical properties used in the equipment. At the cold endof the cycle, the cooling system should ensure a low steam temperature along with therejection of the condensation heat (nevertheless, the lowest condenser temperatures arelimited by the choking phenomenon in the steam turbine: when choking occurs, furtherreductions in exhaust pressure produce no increases in turbine last stage’s work).

The effectiveness of a dry cooling system is most suitably characterised by the overallInitial Temperature Difference (ITD, the „thermodynamic driving force”): the differencebetween the temperature corresponding to the turbine saturated steam back-pressure andthe ambient air temperature. In the case of indirect dry cooling the overall ITD and thetower ITD, ϑt, differ by the value of the condenser Terminal Temperature Difference

INTRODUCTION

16

(TTD). A cooling system cannot be designed for an arbitrarily small overall ITD value,because lower ITD values, i.e. smaller natural draught require greater heat exchangersurface area for a given cooling tower height.

Ambient winds can significantly alter the flow field around a natural draught coolingtower, which will result in an increase in tower ITD as well as in overall ITD values.Consequently, the saturated steam temperature and pressure in the condenser will behigher and the thermal efficiency of the power cycle will be lower which may amount tosignificant economic losses for the respective power company. For instance, if thecooling water temperature in a 1,000 MWe power station is only one degree Centigradehigher due to wind effects, it is estimated that the corresponding decline in thermalefficiency will cause an electric power reduction of around 55 MW (SAMALI AND MADADNIA,2001).

In this Ph.D. work the wind effect on a natural draught Heller type cooling tower willbe analysed by means of on-site measurements and Computational Fluid Dynamics(CFD) calculations. Some possible solutions for improving the cooling towerperformance in wind will also be examined by CFD simulations. The results of thesesurveys may contribute to the design of more efficient cooling towers for the future andthe behaviour of already existing towers in wind may also be improved at a relativelysmall expense.

In this chapter, a brief introduction highlighted the role of power plant coolingsystems, gave an overview of different cooling options, described the operation ofcooling systems employing natural draught cooling towers and explained the importanceof present research and development study. The second chapter summarises the existingexperiences of other researchers in this field which can be found in the scientificliterature. Afterwards, full-scale measurements on a Heller type dry cooling tower arepresented. In the fourth chapter, particulars of the CFD model of the cooling tower aregiven, and the achievements of numerical simulations are outlined in the subsequentchapter. Finally, the conclusions are summarised in the sixth chapter. Photos taken at theon-site measurements are set out in the Appendix.

LITERATURE OVERVIEW

17

2. LITERATURE OVERVIEW

The problem of the influence of wind on natural draught cooling towers had beeninvestigated by several researchers. The methods include full-scale site measurements,scale model tests and numerical simulations.

2.1. Full-scale experience

As a supplier of power plant cooling systems, EGI - Contracting Engineering Co. Ltd. hasmore decades of experiences with natural draught cooling towers. PÁLFALVI (1989)presented results of measurements on two Heller type dry cooling towers in China. Thewell-tried technical solutions applied at the cooling systems for the Hungarian Mátra(former name Gagarin, commissioned in 1971-1972) thermal power station with265 MWth heat rejections were modified slightly only in few aspects for the Units V andVI of the Datong power station in China (commissioned in 1987-1988). In considerationof the significantly high site elevation, the cooling towers in Datong were designed with alittle bit higher structural height and cooling water flow rate as compared with the Mátracooling towers, but the main equipment and the size of exchange area remainedunchanged.

Based on Performance Guarantee Tests carried out on the dry cooling towers ofDatong power station in China, PÁLFALVI (1989) concluded that cooling towers of similardesign behave similarly in cross-wind (see Figure 2-1.).

Figure 2-1. Wind effect curveSource: PÁLFALVI (1989)

Notes: Curve measured in the Mátra PS, Hungary• Datong PS, Unit V+ Datong PS, Unit VIΔ ITD = ITD - ITD0

Δ ITD / ITD0

5 10

0.3

0.2

0.1

0

Wind velocity at 2 m level, m/s

LITERATURE OVERVIEW

18

The experiences with other realised cooling towers showed that higher towers are lesssensitive to wind than lower towers with larger diameter, because of the lower air exitvelocity from the cooling tower in the latter case. Dry tower designers often considerlarger air velocity at the tower exit to reduce the wind sensitivity of the cooling system.On the other hand, a cooling tower with larger bottom diameter has more space availablefor the heat exchangers, with lower air flow velocity and pressure loss through the tower,so the ratio of height to diameter of the towers should be optimised. Generally, naturaldraught dry cooling towers operate with considerably larger draught (larger airtemperature/density difference between tower shell inside and outside), and also theirupper exit air flow rate is 3-4 times larger than that of a wet cooling tower, thus drytowers form stronger vertical outlet air sheaf and better withstanding against wind effects.

In case of fan-assisted natural draught cooling towers the effect of wind on the stabilityof the hot air plume exiting the tower is recommended to be improved by a slightly rota-ted position of the fans, which induces a swirling flow inside the tower (see Figure 2-2.).

Figure 2-2. Example of a fan-assisted natural draught cooling tower

Other two company leaders at EGI, BALOGH AND TAKÁCS (2001) reported results of fieldmeasurements at Bursa CCPP concerning the cooling tower heat rejection capacity fluc-tuations triggered by atmospheric wind-gusts. They judged the heat exchanger air inletflow pattern indirectly from the heat exchanger outlet water temperature measurementsaround the tower perimeter (measurement made on July 2, 2001, at Unit 1 cooling tower).

Heat exchangers in the frontal area, facing directly the wind upstream, hadsignificantly enhanced cooling by utilising their increased air inflow. On the West side(the free side opposite to the side at the power block buildings) the wind flow (tangentialto the perimeter) caused a low pressure in that region and consequently reduced the airflow of the respective heat exchangers. Also the heat exchangers on the power block side(East) had good inlet air flow, since here the buildings blocked and directed cold air tothese coolers. At the backside of the cooling tower also an improved air inflow could bemeasured, due to the refluxes. Thus heat exchanger outlet water temperature variationcurve shows that with vertical heat exchanger arrangement a significant portion of the

LITERATURE OVERVIEW

19

coolers is supported by wind with better cooling, and nearly same heat exchanger area isaffected positively, as the negatively affected heat exchanger area (see Figure 2-3.).

average ofmeasured valuesvalues belowaveragevalues aboveaverage

Cooling tower perimeter

Wind direction

-5 oC 0 oC

+5 oC

Figure 2-3. Outlet water temperatures averaged for every two neighbouring heat exchanger modules(measured values for 4.8 m/s wind speed at 10 m level)

Source: BALOGH AND TAKÁCS (2001)

The wind speed varied in a relatively wide range from second to second, as wind-gustsoccurred. Due to the considerable water fill of the cooling system representing a certainthermal inertia, and due to the actual cooling water pipe lengths, strongly filtered watertemperature and turbine back pressure variation was resulted. This may ensure morecomfortable operating conditions for turbine blading, especially as compared to direct aircooled condensing plants not utilising water as indirect cooling medium.

(Bursa CCPP, Unit 1, July 1, 2001)

0

4

8

12

16

20

0:01

0:32

1:03

1:34

2:06

2:37

3:08

3:39

4:11

4:42

5:13

5:44

6:15

6:47

7:18

7:49

8:20

8:52

Time

Coo

ling

Plan

t Cap

acity

, M

W/K

0

3

6

9

12

15W

ind

spee

d, m

/s

Cooling Plant Capacity

Wind Speed

Figure 2-4. Sharp wind-gusts are filtered by the thermal inertia of cooling water in Heller systemSource: BALOGH AND TAKÁCS (2001)

In Figure 2-4., actual values of Cooling Plant Capacity (CPC) are plotted against theinstantaneous wind speed values measured at ~74 m distance upstream the cooling towerand 10 m from ground level taken in every 20 seconds. CPC is defined by equation (2.1).

a1w1 TTQ−

=CPC (2.1)

No highbuildings

~15-30 m highbuildings

Power plant layout

Wind direction

Bursa CCPP, Unit 1July 1, 2001

Cooling Plant Capacity

Coo

ling

Plan

t Cap

acity

,M

W/K

Win

d sp

eed,

m/s

Time

Wind speed

LITERATURE OVERVIEW

20

CAYTAN AND FABRE (1989) gave results of some on-site tests carried out by Electricité deFrance, and showed the economic importance of wind effect on the performance ofnatural draught wet cooling towers. It was also described how proposals which have thesame nominal specifications can have very different performances over the wholeoperating range if the corresponding cooling towers did not have the same behaviour towind. A method was presented how to choose the best cooling tower design based on thecharacteristics in function of wind speed which is to be supplied by constructors in theirproposal in addition to the nominal parameters.

It was experienced that the cooled water temperature was very sensitive to wind andthat a temperature increase of 3 K can occur at a wind speed of about 10 m/s measured at11 m above ground level. Figure 2-5. gives average cooled water temperature variationsas a function of wind speed at 11 m above ground level, compared to the cooled watertemperature without wind. The cooling water temperature increase due to the wind effectresults in a degradation in turbine performance and loss of electricity production.

Figure 2-5. Effect of wind speed measured at 11 m above ground level on the cold water temperature ofrain counter-flow (left) and cross-flow (right) natural draught wet cooling towers

Source: CAYTAN AND FABRE (1989)

Measurements on these towers were also reported in BOURILLOT AND GRANGE (1980).

For a cooling tower serving a 900 MWe Electricité de France plant, CAYTAN AND FABRE

(1989) estimated the cooled water temperature increase in wind weighted by theoccurrence probability of different wind speed classes to 1.1 K, which corresponds to apower loss in the order of 3 MW or to an actualised financial loss being 19 % of the netinvestment cost. It is therefore worthwhile examining wind effects much more carefully.

To be more competitive, it is in the constructor’s interest to under-estimate the windeffect in his proposal. Therefore, it was advised to allow for wind effect during theperformance test of the tower and apply penalties if the guaranteed wind effect is notfulfilled.

DOSKEMPIROV ET AL. (1992) introduced a construction that consisted of a raw of shieldsfixed along the perimeter at the upper section of the tower with height about 8 % of that

ΔTw2, K

ΔTw2, K

Dampierre 1

Cruas 1Bugey 4

Gardanne 4

5 10v11, m/s

Saint Laurent B1

Emile Huchet 5

4

3

2

1

0 5 10v11, m/s

3

2

1

0

LITERATURE OVERVIEW

21

of the tower (see Figure 2-6.). It provided air flow increase in the cooling tower of about35 %, which was confirmed by the results of experiments on scale model towers and bymeasurements of air flow on two equal industrial cooling towers standing near each otherand having spraying area of 1,200 m2 each. In the latter experiment, only one of thecooling towers had the 5 m height mechanism to increase air flow at wind.

Figure 2-6. Mechanism for air flow increase in cooling tower at wind („wind crown”)Source: DOSKEMPIROV ET AL. (1992)

The influence of wind on tower performance was studied for several years on theNeurat cooling tower in Germany (BAER, 1979), and at the tower outlet a smallrecirculation zone forming a „ring” of cold air inside the tower exit was observed in no-wind conditions. In the case of low cross-wind, the recirculation zone was forming onlyon the windward side of the tower becoming a „wedge” instead of a ring. Thesephenomena were also revealed qualitatively by numerical study in RADOSAVLJEVIĆ AND

SPALDING (1989).

2.2. Scale model laboratory tests

Generally, two kinds of models are used for representing cooling towers in wind tunnel.The first model is called the „cold tower” and is based on an isothermal presentation withforced flow and it characterises the flow pattern and pressure losses in the tower. Thesecond model is called the „hot tower” and is based on the conservation of theArchimedes number, using heated models. The isothermal model implies that thermaleffects due to the air buoyancy are negligible, hence the model employs a fan forproducing tower draughts.

To clarify the mechanism of unfavourable effects of wind on cooling efficiency of drycooling towers, WEI ET AL. (1995) conducted full-scale measurements and wind tunnelmodelling. In the model, a hot water circulation system and finned tube heat exchangerswith the same structure as that of the full-scale tower were used for simulating the

LITERATURE OVERVIEW

22

thermodynamic process of dry cooling towers. A constant water temperature was kept atthe inlet of the heat exchangers and the temperature drop of the cooling water, ΔTw,between the inlet and outlet of the coolers was measured. A dimensionless wind effectcoefficient, Cw, was defined and measured to describe the wind effects on the efficiencyof cooling tower:

w0

w0ww T

TTCΔ

Δ−Δ= , (2.2)

where ΔTw0 is the value of ΔTw when there is no wind. Clearly if Cw is negative, the windhas an unfavourable effect, if Cw is positive, the wind effect is favourable.

In the wind tunnel experiments three different cooling tower models were investigated(see Figure 2-7.):

Model I: a 1:200 scale model of a dry cooling tower in China, in which the heatexchangers had finned tube structure and the same drag coefficient as that of thefull-scale tower, but a different geometry, and were divided equally into six sectors.

Model II: a 1:800 scale cooling tower model without heat exchangers for testing theeffects of lateral wind past the tower exit on the internal flow passing through thetower model. The internal flow was produced by a fan. The static pressure wasmeasured at the internal surface of the throat of the tower model.

Model III: a 1:400 scale cooling tower model with a transparent shell for flowvisualisation employing the smoke wire method. Instead of heat exchangers therewere a copper net and an electric heater at the base of the tower model.

Figure 2-7. Sketch of models I-III (dimensions in mm)Source: WEI ET AL. (1995)

Model I

Model IIModel III

LITERATURE OVERVIEW

23

A boundary layer wind tunnel with a 2.4 m wide, 1.8 m high and 8.0 m long testsection was used. Using spires and roughness elements on the floor, a turbulent boundarylayer simulating the atmospheric boundary layer at the site of the full-scale tower wasobtained. The exponent of the power law of the mean velocity profile was about 0.24.

The experimental results regarding the variation of Cw along the bottom circumferenceof model I can be questionable, because it shows that the heat dissipation has almost thelowest value at the leeward side of the heat exchanger surface. It may be due to that thedrag coefficient of the coolers was too large compared to the other pressure losses in thetower model.

WEI ET AL. (1995) claimed that the value of Cw averaged over the whole heat exchangersurface becomes positive when the ratio of wind speed measured at height correspondingto 10 m at the full-scale to the mean velocity through the heat exchangers exceeds 6.4.From measurements on model II they concluded that the lateral wind can even increasethe tower suction.

MADADNIA ET AL. (1998) described the development of performance improving structureswhich alleviate wind-induced performance losses of a natural draught wet cooling tower.The physical dimensions of a representative 660 MWe natural draught wet cooling towerwere scaled down 1:1,000. An isothermal and decoupled model was constructed and usedin a purpose built wind tunnel. The tests were concentrated on flow effects around theinlet region, since this area is the most promising for the retrofitting of a cooling tower.The curative impact of three effective performance improving structures on the towerperformance was quantified by measuring changes in the dimensionless tower’s inletstagnation pressure loss coefficient, Cpi, defined as

⎟⎟⎠

⎞⎜⎜⎝

⎛=

⋅⋅Δ

=w

t2t

pi vv

vpC fρ5.0

, (2.3)

where Δp is the stagnation pressure loss in the tower (measured between tower outsideand inside pressures at the reference positions),

vt is the reference air velocity in the tower, downstream of the packing,vw is the cross-wind velocity at the reference height.

In their paper, MADADNIA ET AL. (1998) quoted also some estimations for the effect ofcooling tower operation on the power plant efficiency. Typically, in the United Kingdoma reduction of 0.5 °C in cooling water temperature (approximately a fall of 2 kPa incondenser pressure) would save about A$ 300,000 per year in fuel costs on a typical500 MW unit (RENNIE AND HAY, 1990). Assuming the electric power generation capacity ofBritish power stations was around 63 GW, the saving would amount to $ 37.8 Mannually. According to an estimate by the New South Wales Department of Minerals andEnergy there will also be a corresponding reduction of tradeable CO2 emissions by300 tonnes annually in a power station equipped with two typical 660 MW units and twocooling towers in the case of a 0.5 °C reduction in cooling water temperature. It was also

LITERATURE OVERVIEW

24

noted that a 1 °C increase in cooling water temperature could easily occur when the windvelocity, measured at a height of 2 m, changes from 2.4 m/s to 4.4 m/s.

The isothermal model in MADADNIA ET AL. (1998) was fitted with a mechanically drivenfan to provide a uniform stream of air flow inside the shell, simulating no-wind con-ditions. The fan flow rate was controlled by a motor-speed controller (see Figure 2-8.).The reference wind pressure and velocity were measured at a height of 10 mm above thetower’s base level. A Fecheimer probe was used for wind velocity and pressuremeasurements, positioned 50 mm upstream of the tower at the reference height. Inside thetower, a Pitot tube was used for air velocity and pressure measurements. Wind velocity,air velocity at the throat of the tower and stagnation pressure difference between thethroat of the tower and the ambient were measured. Numerous flow conditioning deviceswere tested and refined whilst observing their effect in reducing wind-inducedperformance losses.

Figure 2-8. The scale model in the purpose built wind tunnelSource: MADADNIA ET AL. (1998)

Figure 2-9. Performance curve with and without a curative deviceSource: MADADNIA ET AL. (1998)

Wind directionGround level

Cooling tower model(inverted)

Transition duct Flow straighteningscreens

Induced draught ducting

Centrifugal induceddraught fan

Axial fan

0.05 0.1 0.15 0.2 0.25 0.3vw / vt

1.2

1.0

0.8

0.6

0.4

0.2

0

ma /

ma0

No curative device

Curative device #1

r = 1.5·hi

3·hi

3·hi

LITERATURE OVERVIEW

25

An improvement in tower performance was observed through to both reduction of upto 88 % in Cpi and an increase of up to 118 % in mass flow rate of air through the towerwith rain zone only (no packing) compared to the tower performance with noperformance enhancing structures. In addition, it was noticed that the effectiveness of thecurative devices decreased as packing was placed in the model with increasing resistance.In Figure 2-9., ma is the tower air mass flow rate with wind, ma0 is the tower air massflow rate with nil wind and hi is the tower intake height.

SAMALI AND MADADNIA (2001) presented a boundary layer wind tunnel experiment whichwas used to simulate wind flows for cooling tower studies on a 1:250 scale model of a250 MWe prototype natural draught cooling tower. It was stated that in studiesconcerning the flow pattern around the inlet region of a cooling tower, modelling theimmediate topography is also important. Every obstruction, minor structure, hills andvalleys and other features should be properly modelled to correct scales. This can beachieved by using blocks as roughness elements on the floor of the wind tunnel.Alternatively, as cooling towers are usually located in open terrain featuring lessturbulence, ground effects can be modelled using carpets in addition to mesh and spiresupstream of the working section of the tunnel (see Figure 2-10.).

Figure 2-10. Cooling tower model in the environmental wind tunnelSource: SAMALI AND MADADNIA (2001)

It was mentioned that windscreens, windbreak walls, sound attenuators (i.e. plantedembankments), sound absorption walls and sound absorption louvers etc. can also serveas structures to lessen the effect of cross-wind on tower performance. Structures of thistype and in particular silencers installed directly at the air inlet have a negative effect onthe cooling tower size due to the high pressure losses. In addition, there is an increase incapital cost. Windscreens reduce the effects of cross-winds through equalising the air

LITERATURE OVERVIEW

26

inlet flow and directing the air flow into the packing system as well as help to preventwater spray losses out of the air inlet opening in case of wet cooling towers.

SAMALI AND MADADNIA (2001) fitted their isothermal model with a mechanically drivenaxial fan to provide a uniform stream of air flow through the tower, simulating the wind-off flow conditions. They argued that although matching wind velocity and turbulenceintensity profiles is important, the quality of test results will primarily depend on exactmodelling of immediate topography. Hence the use of nearby artificial obstacles such asblocks and trip boards, used to generate the required velocity and turbulence intensityprofiles, is not recommended in the immediate vicinity of the tower.

ZHAO AND SHI (1998) studied some measures to overcome the negative wind influenceon Heller type dry cooling towers through model test in wind tunnel. Field test for twoHeller type dry cooling towers in Datong Second Power Plant, China, serving two sets of200 MW generating units with 119 cooling deltas and a tower height of 125 m was alsomentioned, carried out by Cooling Water Department of Institute of Water Resources andHydropower Research and Beijing University in May 1991. The observation resultrevealed that the averaged inlet air velocity was about 6 m/s without natural wind anddecreased to 4.8 m/s with the natural wind velocity of 5 ~ 7 m/s, whence the air tempera-ture inside the tower increased by about 7.5 ºC.

Figure 2-11. Variation of air velocity into the tower with tower top wind velocitySource: ZHAO AND SHI (1998)

Since ZHAO AND SHI found only few data in the literature regarding the effect of wind ondry cooling towers, they considered that the external configuration of a dry cooling toweris about the same as that of a wet cross-flow cooling tower, and utilised some test datafrom the latter. A cross-flow natural draught wet cooling tower equipped for a 900 MWgeneration unit has been chosen as the tested prototype. It was 125 m in height, 175 m inbase diameter, and the air flow velocity at packing section was vi0 = 2.2 m/s. Beside fieldmeasurement, model tests with scale ratio of 1:175 were also analysed: it was indicatedthat the inlet air flow velocity around the tower is positive under all natural wind velocityconditions. The most unfavourable region with the smallest air inlet velocity undernatural wind condition was at the leeward and side part of the tower. The variation of themean air flow velocity into the tower, vi, with the natural wind velocity, v∞, is shown inFigure 2-11., which indicates that only strong wind (the case of v∞ / vi0 > 11.7 on the

v i / v

i0

1.1

1.0

0.9

model

prototype

v∞ / vi0

0 5 10

LITERATURE OVERVIEW

27

model) is good for cooling. The effect of ordinary natural wind is alwaysdisadvantageous.

Model study for further investigation of wind effect was conducted for another Hellertype dry cooling tower with tower height of 115 m, base diameter of 97.2 m, outletdiameter of 58.6 m, number of cooling deltas 107, height of heat exchangers 15 m, airinlet height of heat exchanger 16.6 m, height of cross supporting columns of the tower18.5 m, water mass flow rate 6,111 kg/s, water temperature difference into and out oftower 12 ºC, air mass flow rate 13,375 kg/s, environmental temperature 26.5 ºC, airtemperature difference inside and outside the tower 22.9 ºC, inlet air flow resistancecoefficient 21 (including the resistance of blind window, heat exchanger, cross columnand intake screen). Undistorted model with length scale 1:180 was used. Strict geometricsimilarity of heat exchanger and blind window of the tower was replaced by the equalityof the air intake resistance coefficients of the model and the prototype.

The cylinder of the model tower was made by glass fibre reinforced plastic plate 3 mmthick. There were 6 sectors of heat exchangers, being the same as that in the prototype.Each sector consisted of 12 copper tubes of 7.6 cm in length, 1.0 cm in inner diameterand covered by 0.3 mm thick plate fins with 2.7 mm spacing between them. Each set hadits own inlet and outlet pipes. The water flow rate and water temperature at the inlet ofthe six sectors were kept to be the same and the water temperatures at the outlets weredifferent under different wind conditions. The water temperature difference of the inletand outlet of each heat exchanger was used to indicate the effectiveness of thecorresponding sector. The model test was conducted in a wind tunnel with a 2.4 m wide,1.8 m high and 8 m long test chamber. The water temperature at each inlet of the heatexchanger was taken as ~ 100 ºC and the temperature difference between inlet and outletwas kept to about 12 ºC under no-wind condition.

Figure 2-12. Effect of wind on water temperature, Figure 2-13. Effect of wind on water temperature,original arrangement (no guide-walls) arrangement with the suggested improved measure

Source: ZHAO AND SHI (1998)

The variation of the inlet-outlet water temperature difference with different windvelocity is shown in Figure 2-12., which indicates that the most unfavourable effect on

0 1 2 3 4 v∞, m/s

6

4

2

0

-2

-4

-6

-8

ΔTw - ΔT

w0,

ºC

ΔTw - ΔT

w0,

ºC

0 1 2 3 4v∞, m/s

10

8

6

4

2

0

-2

LITERATURE OVERVIEW

28

the cooling tower occurred at 1.4 m/s wind velocity which corresponds to the prototypewind velocity of 9.3 m/s at 10 m elevation above the ground.

The cooling ability of the tower was reduced under wind condition, the temperaturedrop of all sectors, except the upwind sector, decreased. The reduction of the heatdissipation from the tower was 31,300·mw (W), where mw (kg/s) is the cooling waterdischarge for all heat exchanger sectors.

Four guide-walls were added outside the air inlet around the tower periphery assketched in Figure 2-14., of which two guide-walls were set along with the winddirection and the other two normal to it. Two sizes of guide-walls were used, one was16.5 m high (same height as the heat exchanger), 10 m wide, and the other was 14 mhigh, 10 m wide. The heat dissipation of the tower for the case with 16.5 m × 10 m wallswas increased by the amount of 4,600·mw (W).

With the guide-walls added, the air flow distribution into the tower remainedunchanged without wind, but the air flow rate in both the upwind and lee sides increasedwell under wind condition. The worst fluid area was shifted to out of the two side guide-walls. Thus the flow pattern of the air entrance to the tower and thereby the tower’scooling efficiency were greatly improved. Experiments on the tower with more guide-walls were also conducted. No better but worse effect was observed.

Figure 2-14. Figure 2-15. Figure 2-16.

Guide-walls outside the tower Gateway on guide-walls Cross-walls inside the towerSource: ZHAO AND SHI (1998)

Experiments with natural wind of 15 º, 30 º and 45 º inclinations have led to similarconclusions. The variation of the tower temperature difference with the wind velocity isshown in Figure 2-13.

A 3 m high and 2 m wide gateway was provided in the above mentioned case with16.5 m × 10 m guide-walls (see Figure 2-15.) so as to furnish a traffic way around thetower and reduce the wind pressure upon the guide-walls. The gateway showed no greatinfluence to the tower. The improvement of this case was something like that of case with14 m × 10 m guide-walls. Gateway of large size (3 m × 8 m gateway in 14 m × 10 mguide-walls) led to rather undesirable results.

Gateway Guide-wallGuide-wall

LITERATURE OVERVIEW

29

Cross-walls inside the tower of 18 m height and 63 m length were also investigatedwith one wall set in the wind direction and the other normal to it (see Figure 2-16.). Nosignificant improvement was obtained by using such arrangement.

An added cone of 34 m height and 63 m base diameter (see Figure 2-17.), with theguide-walls set in the same way as in Figure 2-16. inside the tower also yielded an effectinferior to that of cases by adding guide-walls outside the tower.

Figure 2-17. Cone and cross-walls inside the tower Figure 2-18. Wind gathering chamberSource: ZHAO AND SHI (1998)

A 180 º wind gathering chamber of 16.5 m height and 10 m width on the lee side of thetower (see Figure 2-18.) resulted in an increased total heat dissipation, by 51,500·mw (W)more than that in the case of no wind. The shortage of this measure is its high engineeringcost, and it is good only for a specified wind direction.

There could be taken some improvement measures also on top part of the tower, butthey were found to be less effective and more expensive in capital cost due to theadditional weight on the tower top. Adding four guide-walls outside the air inlet (Figure2-14.) was finally chosen as the suggested arrangement.

Two dry cooling towers of the concerned power plant have been installed as sketchedin Figure 2-19. A 8 m high wall set aside 10 m from the towers and a 12 m × 20 m × 8 mhouse is located between the towers. Experiments of the said arrangement with thesuggested improvement measure and that of the original one were performed and theimprovement effect was even more satisfied than that in the single tower experiment.

Figure 2-19. Engineering arrangement experimentSource: ZHAO AND SHI (1998)

Interfering tower

1.4 m/s

Wall

Test tower

LITERATURE OVERVIEW

30

Earlier EGI also carried out wind tunnel tests (CSABA, 2003). First, a pipe of finitelength was closed off at one end and placed in the wind tunnel. It was pointed out that inthe free end of the pipe a depression nearly equal to the dynamic pressure of wind wasinduced when the pipe was subjected to a perpendicular air flow. Later on, the closed endof the pipe was opened and a filled inlet was mounted onto it, this way ensuring that thedepression could suck in the air at this end of the pipe, so the air began to flow throughthe pipe. The filled end of the pipe was connected to a square-shaped flat plate so that theplate was perpendicular to the axis of the pipe. The air velocity inside the pipe wasproportional to the wind velocity. It was observed that when the air flow was throttled inthe pipe, then the induced depression caused lower air velocities through the pipe thanwithout throttling. Conversely, when the air flow through the pipe was increased abovethe equilibrium value by a fan, then an additional pressure loss occurred, which tried todecrease the air velocity inside the pipe to the equilibrium value.

It was therefore stated that if the air exit velocity related to the wind speed is lower, theeffect of wind is more favourable at the tower outlet. However, it should be emphasisedthat for the appropriate stability of the air outflow and due to economic considerations,criteria that are opposite to the above requirement should be met. It was also claimed thatthe deteriorating effect of wind is not predominant at the tower outlet.

LUCAS AND BUCHLIN (1986) investigated cold air inflow using a vertical heated tube (seeFigure 2-20.). It was pointed out that the entrainment of cold air occurs when the netinertia of the plume is unable to support the overlaying heavier air. This was the case ifthe inlet pressure loss coefficient, Cpi (defined by equation (2.3) and varied by changingthe inlet gap of the cooling tower model) exceeded 35. The motion of heated air near thecentre of the test section had two types: pure laminar flow for Cpi less than 40 and puffingat higher values of Cpi. During puffing the cold air was sinking into the central portion ofthe cylinder and due to a shearing force between the upward flowing heated air and thedescending cold air, a counter rotating vortex was produced. The penetrating vortexcontinued to mix until the buoyancy was too small to carry the puff against theupcurrents. The vortex stalled at some position inside the cylinder and then wasexhausted from the test section.

Figure 2-20. Entrainment of cold air into the centre of the test section (puffing)

Flow

LITERATURE OVERVIEW

31

The inflow of cold air near the crown of a cooling tower (which leads to a negativedraught force) has been attributed to a decrease in the cooling tower performance. Theflow is attached when the momentum of the exhausting fluid maintains a shear layerpreventing the inflow of cold air. The flow is strongly separated if the mixing of thefluids is unable to immediately exhaust the penetrating cold air. In general, separationoccurs randomly, the motion is erratic and does not remain at one point along the rim.

RICHTER (1969) and BUCHLIN AND OLIVARI (1982) suggested that the instabilities begin atan Archimedes number (defined by equation (2.4), where D is the diameter of testsection) greater than 3.

2vDgAr

⋅⋅Δ⋅

=ρρ (2.4)

2.3. Numerical simulations

At different wind speeds, the three-dimensional and turbulent flow around and inside acooling tower is extremely complex. Various vortex systems are set up which interactwith each other. A predictive understanding of the flow around cooling towers is usuallyobtained from model studies carried out in wind or water tunnels. Such studies arelaborious, time-consuming and expensive. Hence, there is a need for reliable mathema-tical models with which parameter studies can be carried out. Numerical methods areusually free from many of the practical constraints imposed on the prototype and scalemodel experimental methods and might provide additional valuable information.

The quantitative simulation of dry and wet cooling tower performance have been beingapplied since many decades ago. Early theories, for example those of CHILTON (1952) andRISH (1961), treated the flow through the tower as one-dimensional, and as governedsolely by the buoyancy of the air, the pressure losses in the components of the tower, andthe heat-, and in case of a wet cooling tower, the mass transfer processes between the airand the water. The above theories also made use of Merkel’s approximation (MERKEL,1926) for the evaluation of the „number of transfer units”. Later theories (POPPE, 1972 andMESAROVIĆ, 1973) introduced various improvements, but none modelled both internal andexternal aerodynamics simultaneously. RADOSAVLJEVIĆ AND SPALDING (1989) investigated thewind influence on large wet natural draught cooling towers of the type used in manysteam power stations for cooling large quantities of water by direct contact with theatmosphere. They used PHOENICS (SPALDING, 1981), a general purpose computer codefor fluid flow simulation to obtain numerical solutions of the three-dimensional thermaland aerodynamic flow fields inside and outside the tower, by numerical integration of theheat and mass transfer equations.

Their steady state and single phase simulation of the tower performance under theprescribed conditions predicted outlet water temperature as well as enthalpy, vapourconcentration, velocity and pressure fields inside and outside the tower. Heat and mass

LITERATURE OVERVIEW

32

exchange between water and air, and resistances to the flow inside the tower weremodelled by source terms in the enthalpy, concentration and momentum equations,respectively. The calculations were performed within a three-dimensional spatial domain(two-dimensional for no-wind case), described by a non-uniform curvilinear grid alongthe radial, tangential and vertical cylindrical coordinate axes r, ϕ and z, divided into 26, 6and 42 cells, respectively. The solution domain was a half cylinder 300 m high with200 m radius. Tower shape was defined by the so-called body-fitted grid, which followsexactly the hyperbolic shape of the shell. The investigated wet cooling tower wasassumed to serve a power plant unit of 350 MW, the cooling water flow rate was8.556 m3/s, the tower height was 115.4 m, the radii at the base, throat and exit of thetower were 42.6 m, 22.8 m and 24 m, respectively. Calculations have shown that furtherwind velocity increase above 1.5 m/s has positive effect on the tower performance,decreasing the outlet water temperature. It was also noticed that the warmest region of airand water was shifted towards the windward side of the tower when subjected to cross-wind. Calculations for a different tower geometry (Tower 2 with an exit radius bigger by1.5 m) were also performed to show its influence on the performance. While in the casewithout cross-wind both towers showed similar performance in respect of water outlettemperature, Tower 2 performed significantly worse in the case of low cross-wind,because of a deep penetration of cold air down the tower, decreasing the effectivechimney height. Further increase in wind velocity brought the same decrease in wateroutlet temperature as for Tower 1, but the overall response of Tower 1 to wind influencewas much smoother (see Figure 2-21.).

Figure 2-21. Outlet water temperatures in function of the wind speedSource: RADOSAVLJEVIĆ AND SPALDING (1989)

More details about PHOENICS computer code can be found in ROSTEN AND SPALDING

(1987a, 1987b, 1987c).

Using the finite volume method (FVM), SU ET AL. (1999) developed a software calledNumerical Simulation in Turbulence Research during 1991-1994, and simulatednumerically the fluid flow and temperature distribution around and inside a Heller typedry cooling tower under cross-wind. The model equations were built up for steady andthree-dimensional flow of an incompressible fluid. The Boussinesq eddy-viscosity

30

29

28

27

26

25 0 3 6 9 12 15

v∞, m/s

T w2,

ºC Tower 1 Tower 2

LITERATURE OVERVIEW

33

hypothesis was used for Reynolds stress. When air flows through the heat exchangers, aheat source term was added in the energy equation:

( )aw TTAQ −⋅⋅= α , (2.5)

where α denotes the heat transfer coefficient and it was determined from experience(ROHSENOW, 1985a), wT is the mean temperature of water in the heat exchanger and Ta isthe air temperature. The used grid system was non-orthogonal and non-staggered, theapplied discretisation methods were the Rhie and Chow’s momentum interpolationmethod (RHIE AND CHOW, 1983), the Hybrid Linear/Parabolic Approximation (ZHU, 1991)and for the coupling of velocity and pressure fields the SIMPLEC method (PATANKAR,1980) was employed. The linear algebraic system of equations was solved by means ofStrong Implicity Procedure (STONE, 1968).

Due to the symmetry of the fluid flow, only half of the flow field was simulated. At theentry, the velocity, temperature, turbulent kinetic energy and turbulent dissipation ratewere given according to the atmospheric data in Datong. Downstream, the variation of allflow field values along the main wind direction were eliminated. Because of the capacityand speed limit of the computer (PC 486 with 16 Mb memory), the grid number was 36 ×22 × 30, i.e. 36 grid points in the main flow direction, 22 grid points along thecircumference of the tower and 30 grid points in the vertical direction. The heatexchanger zone in the cooling tower was divided into 21 parts along the circumference.One can assume that every heat exchanger section has one entry and one exit for water.The water temperature Tw1 at the entry of the heat exchanger was given, the watertemperature Tw2(j) at the exit of the jth heat exchanger section can be calculated from theheat balance between water and air:

[ ] [ ](j)TTc(j)mT(j)TAQ(j) w2w1wwk

awk −⋅⋅=−⋅⋅= ∑α , (2.6)

where j is the number of sections of the heat exchanger zone, k is the number of gridpoints in the vertical direction, (j)Tw is the mean water temperature in the jth section. It

was assumed that the temperature distribution in the vertical direction is uniform, so( )(j)TT(j)T w2w1w +⋅= 2

1 . mw(j) is the water mass flow rate in the jth heat exchanger section

and cw is a flux coefficient determined from experience (ROHSENOW, 1985b).

The calculations were performed without cross-wind and with wind speeds of 5 and10 m/s. In cases with different wind speeds, the warm water temperature was heldconstant and various heat dissipation values were obtained. However, in a real coolingtower a nearly constant heat load should be dissipated, and it means that the warm watertemperature will increase with the cross-wind in order to reject the heat load remainingalmost unchanged. Furthermore, in this model the influence of the zig-zag cooler position(called „delta” arrangement), which is typical in the Heller system, is neglected, and themodel treats the cooling water is flowing in a simple cross flow relative to the air flowinstead of the cross-counter flow applied commonly in the Heller system.

LITERATURE OVERVIEW

34

The numerical results were compared with full-scale measurement data. The numericalresults were checked also with a finer grid system, where the grid number was 70 × 42 ×60. The influence of cross-wind speed on the temperature of the cooling water isdisplayed in Figure 2-22. The solid lines denote the measurement results of some powerstations, and the dotted line is the result by SU ET AL. (1999).

V∞

Figure 2-22. Influence of wind speed (measured at 10 m above ground) on outlet water temperature1: Datong PS, China, Q = 280 MW, ITD = 28 °C; 2: Ibbenbüren PS, Germany, Q = 188 MW, ITD = 28 °C;3: Razdan PS, USSR, Q = 280 MW, ITD = 30 °C; 4: Gagarin PS, Hungary, Q = 265 MW, ITD = 26 °C;5: Grootvlei PS, Unit 5, South Africa, Q = 331 MW, ITD = 45.6 °C (considerably high ITD!);6: Grootvlei PS, Unit 6, South Africa, Q = 331 MW, ITD = 45.8 °C (considerably high ITD!).

Source: SU ET AL. (1999)

ERNST AND WINKLER (1994) showed how well the flow at cooling towers can be predictedwith the mathematical model MIMO, which solves on a numerical grid the conservationequations of mass, momentum and energy and, if necessary, transport equations forhumidity, liquid water content and a passive pollutant. The model, which was developedon the base of the meso-scale model of FLASSAK AND MOUSSIOPOULOS (1987), works with thewell-known k-ε model of turbulence (LAUNDER AND SPALDING, 1974; RODI, 1980).

One configuration investigated by ERNST AND WINKLER (1994) consisted of somerectangular wet cell cooling towers with a cross-section of 70 m × 30 m and a height of20 m arranged in the shape of the capital letter L with a distance of 100 m between theircentres. A constant wind speed of 7.5 m/s above a height of 100 m was assumed and upto this height the power law with an exponent of 0.3 was used. The thermal stratificationof the atmosphere was slightly stable with a temperature gradient of -7 K/km. Therelative humidity of the ambient air was assumed as 90 % independent of the height.Only the visualised flow field and plume dispersion were discussed in their paper.

Figure 2-23. shows the flow of the plumes in vertical cross-sections perpendicular tothe wind direction along the direction of wind. At the beginning two double vortices areformed with their smaller parts inside. In the course of dispersion, both double vorticescome closer together associated with a rise of the inside vortices. Subsequently, the insidevortices are dissolved and only one double vortex remains.

The vortices can be observed at the outlet of a natural draught cooling tower, too. Thetypical height of a natural draught cooling tower is 100-150 m, and therefore the inlet and

76

54

3

21

0 2 4 6 8 10 12 14 16v∞, m/s

T w2 -

Tw

20, º

C1

2

3

4

56

LITERATURE OVERVIEW

35

outlet openings for the air flow are distant enough to each other to avoid the hot air to re-enter the tower. However, as a mechanical draught cooling tower is much more lowerthan a natural draught one, the inlet and outlet openings are close to each other and thedanger of hot air recirculation is present, which can deteriorate the cooling efficiency.The vortices caused by wind can even result in trans-circulation, when the hot air exitingthe cooling cell flows to the air intake of a gas turbine unit, and thereby reduces thepower of the gas turbine (SZABÓ, 2002).

Figure 2-23. Dispersion of the plumes of two lines of cell cooling towers in vertical cross-sectionsperpendicular to the wind direction

Source: ERNST AND WINKLER (1994)

RAZAFINDRAKOTO ET AL. (1998) described the governing equations for air and water flowsas well as the numerical algorithms used in EdF’s finite element code N3S-AERO.Validations were also carried out by other researchers from EdF on some configurationsof cooling towers in 2D and 3D geometries, showing the ability of N3S-AERO to predictmajor physical phenomena in cooling towers. Global results on thermal performancewere compared with calculations by EdF’s one-dimensional qualified design codeTEFERI (BOURILLOT, 1983; GRANGE, 1994) or experimental data when available.

Among others, a 3D simulation of the wet cooling tower of St. Laurent with crossingwind was performed (FOURNIER AND BOYER, 2001). The cooling water mass flow rate, thebase diameter and the height of the tower were 33,737 kg/s, 175 m and 125 m,respectively. Because of the symmetry of the flow, only half of the domain was includedinto the computation. Inlet flow was given by a vertical power law profile of wind

Wind direction

Cell cooling towers

LITERATURE OVERVIEW

36

velocity with an exponent of 0.2. The size of the computational domain was 2,000 m ×1,000 m × 1,000 m. The mesh consisted of 118,374 elements.

Figure 2-24. shows a typical flow pattern of a cross-flow tower at St. Laurentcalculated by N3S-AERO with crossing wind of 9 m/s at 11 m height above ground. Onepart of air motion turns around the tower as flow pattern around a cylinder. Another partis draught into the tower at its base and creeps up to the top of the tower. Two counterrotating vortices can be observed in the tower. At the exit of the tower, air flow has arolled motion. The stream lines are coloured by air temperature. Water temperature in theexchange zone and plume are also represented. The averaged cold water temperature in-creased by 3.9 ºC and the air mass flow rate through the tower decreased by 41.45 % dueto cross-wind compared to the results from 2D axi-symmetrical calculation without wind.

Figure 2-24. St. Laurent cooling tower with cross flow exchange configuration, 3D calculation with windSource: FOURNIER AND BOYER (2001)

BANQUET AND DELABRIERE (1998) outlined the utility of the N3S-AERO 3D code andsome of its industrial applications, which complements the design and test facilities usedat EdF. It was pointed out that a 3D design code is able to characterise the influence ofwind on performance and research into systems to limit its effects (inlet air guides). Theengineering time required for a complete 3D study with N3S-AERO was roughly 2 to 3months, including 2 weeks to construct the mesh. For a 3D calculation on a CRAY C98computer, results processing to convergence took about 90 hours with a mesh elementnumber of 22,549 in case of a natural draught counter-flow wet cooling tower, and cc.260 hours with a mesh element number of 99,665 for a forced mechanical draughtcounter-flow cooling cell.

DREYER ET AL. (1998) found from measured test data that the performance of two naturaldraught wet process cooling towers at a petrochemical plant in South Africa did notdecline with increasing wind speed as was expected, but an increase in thermal

LITERATURE OVERVIEW

37

performance was detected at high wind velocities. (Wind speed was measured atapproximately 1.5 m from ground level on the windward side of each cooling tower,using a Schildknecht propeller type anemometer.) A CFD analysis was done for thesetowers with firstly the splash pack hung completely within the shell of the tower (Case 1),and secondly with the splash pack hung well into the rain zone (Case 2). A tower with thesplash pack hung into the rain zone was a lot less sensitive to wind than a tower with thesplash pack located completely inside the shell.

Figure 2-25. Section through a typical natural draught cooling towerSource: DREYER ET AL. (1998)

The total height of the full-scale tower was 150.84 m above the pond and the air inletheight was approximately 11.4 m. The shell diameter at the tower inlet lip and toweroutlet was approximately 105.4 m and 65.5 m, respectively. The towers were fitted withplastic splash grids spaced at approximately 1 m intervals. The packing hung into theopen area below the lip of the towers (see Figure 2-25.). The average packing depth was12 m. The rain zone height below the packing was approximately 4 m. Each tower wasfitted with twenty small wind-walls spaced around the perimeter of the tower. Theyextended from the pond level up to the height of the shell inlet lip. At the pond level, thewind-walls extended approximately 7 m into the tower. Their main function was toreduce blow-out of the water droplets from the perimeter of the tower.

The commercial CFD software code, STAR-CD, was used for solving the Navier-Stokes equations using a finite volume discretisation method and solving the finalequations in an iterative manner. The numerical mesh consisted of approximately500,000 mesh cells with a symmetry plane boundary condition and embedded meshrefinement in the region of the tower and packing.

A power law wind speed profile was used as an inlet boundary condition (upstream ofthe tower, with free stream boundaries at the top and side boundaries) and a zero gradientoutlet boundary was applied at the outlet of the domain downstream of the tower. Thesplash packing was simulated using a porous media, which imposed a similar pressure

Air outlet

Tower shell

Drift eliminatorWater distribution

Packing material

Air inlet

Water outlet

LITERATURE OVERVIEW

38

resistance as would the packing. The simulation model was also set up to solve for thevapour concentration in the air. The coefficients for heat and mass transfer were obtainedfrom empirical data for the given splash packing configurations that were simulated. Themass of the water that evaporated in each finite volume cell was injected into the fluidstream by means of a water vapour source term. In this manner the water vapourconcentration was increased to a point where the air becomes saturated.

A summary of the air mass flow data is presented in Figure 2-26. From this graph itcan be seen that the tower with the packing hanging into the air inlet below the edge ofthe shell showed an increase in the air mass flow rate for wind speeds of between 8 and10 m/s. The CFD analysis on the effect of the position of the packing material in thenatural draught wet cooling tower confirmed the findings of the full-scale testsperformed. However, it should be noted that the positive effects of wind could also beattributed to the presence of the twenty small wind-walls.

Figure 2-26. Air mass flow rate (tower draught) versus wind speedSource: DREYER ET AL. (1998)

DU PREEZ AND KRÖGER (1993) investigated the influence of cross-winds on theperformance of a natural draught dry cooling tower by means of full-scale measurementsas continuation of their work on a scale model from 1992 (DU PREEZ AND KRÖGER, 1992).By monitoring all the relevant independent variables, the results of these measurementsallowed direct comparison with the wind effect predicted by a numerical procedureemploying the PHOENICS code. It was shown that results obtained on full-scale towerscan be successfully reproduced provided that the effects of all tower components areincluded in the analysis. In practical cooling towers rectangular heat exchanger bundlesare arranged either vertically around the circumference of the tower or in the horizontalinlet cross-section of the tower and the wind effect is found to be dependent on theparticular layout.

The investigated concrete cooling tower was 165 m high with an inlet diameter andinlet height of 144.5 m and 24.5 m, respectively. The inlet shape of the hyperbolic towershell had a cone apex angle of approximately 24 º while the outlet was essentiallycylindrical. The finned tube heat exchangers, in the form of A-frames with an apex angleof 60 º, were arranged radially in the horizontal inlet cross-section of the tower.

1.00

0.75

0.50

0.25

0.00

ma /

ma0

Case 1 Case 2

1 5 10v∞, m/s

LITERATURE OVERVIEW

39

The results suggested that as the heat rejection rate in a cooling tower increases, thetower becomes less sensitive to cross-winds. This is in agreement with similarmeasurements performed by MARKÓCZY AND STÄMPFLI (1977) on a tower with the A-framesarranged vertically around the circumference of the tower. The mean air mass flow ratethrough the tower was obtained by integrating the readings of the 12 anemometerslocated in the throat of the tower. The results of these measurements were presented asthe ratio of the air mass flow rate with wind to the mass flow rate obtained in windlessconditions (see Figure 2-27.).

Figure 2-27. Reduction in the mean air mass flow rate Figure 2-28. Grid distribution,through the tower due to the wind an array of A-frames

Source: DU PREEZ AND KRÖGER (1993)

The effect of the A-frame arrangement was taken into account by employingcorrelations for the temperature and pressure loss coefficient correction factor for an A-frame in function of the incidence angle, instead of modelling the exact geometricalconfiguration. These correlations were determined for two-dimensional A-frames with anapex angle of 60 º by using the PHOENICS computer code (see Figure 2-28.). Thepressure loss coefficient for normal flow through the individual heat exchanger bundleswas given by an empirical expression. As a result of the inclined flow, it was observedthat as the flow incidence angle to the A-frames decreased, the flow distribution throughthe A-frame became increasingly more non-uniform, while the flow rate through theupstream heat exchanger within an A-frame decreased (due to an increase in the effectivepressure loss coefficient) and the mass mean temperature of the air leaving the coolerswas less than that which would be found for a uniform velocity distribution.

The influence of the heat exchanger arrangement on the wind effect on a dry coolingtower was also investigated by means of the numerical model. The same tower geometrywas used in the calculations but the A-frames previously arranged horizontally andradially in the tower, were positioned vertically around the circumference of the towerdirectly below the inlet edge of the tower shell. The apex angles of the A-frames werekept at 60 º, but the length of the fin tubes, and therefore the inlet height of the tower,were changed to retain the total frontal area of the heat exchangers. With a tube length of

Heat rejected by tower = 650 MW

• Full-scale measurementsHorizontal A-frame arrangementVertical A-frame arrangement

ma /

ma0

1.05

1.00

0.95

0.90

0.85

0.80

0.75

0.700 1 2 3 4 5 6 7 8 9 10 11

v96, m/s

xz

LITERATURE OVERVIEW

40

19 m the inlet diameter to inlet height ratio of the tower was found to be 7.7 incomparison with the 5.89 of the previous model.

By applying a CFD technique using the general purpose commercial FLUENT code,AL-WAKED ET AL. (2001) investigated the effect of cross-wind on the thermal performanceof a natural draught wet cooling tower. A steady state three-dimensional study using thestandard k-ε turbulence model to simulate incompressible air flow in and around thecooling tower has been conducted. The code has been validated against the available datain the literature. A case study has also been conducted to examine the effect ofintroducing curative devices at the lower part of the cooling tower. The introduction ofthe curative device has indicated an improvement in alleviating the thermal performancelosses due to cross-wind.

The heat rejection was modelled using the radiator boundary condition in FLUENT:

( )H2HX TTq −⋅= α , (2.7)

where q is the heat flux, α is the convective heat transfer coefficient, THX is the heatexchanger temperature and TH2 is the exit fluid temperature. The resistances of the heatexchangers to the air flow were presented as sinks in the momentum equations, and theflow through the radiator has been assumed to be vertically only. This was modelledusing the porous media boundary condition.

The simulations were performed for the following set of conditions:

Dry bulb air temperature: 12 ºC Block power: 200 MWth

Wet bulb air temperature: 7 ºC Tower height: 130 m

Mean radiator temperature: 40 ºC Base radius: 50 m

Atmospheric pressure: 87 kPa Throat radius: 25 m

Wind-wall height/width: 10 m/10 m Exit radius: 28 m

Figure 2-29. General view of the Figure 2-30. Effectiveness of the cooling towercooling tower with wind-walls with and without wind-walls under cross-winds

Source: AL-WAKED ET AL. (2001)

0 5 10 15 20 25v∞, m/s

Toweroutlet

Towerinlet

Shell

Wind-wall

120

100

80

60

40

20

0

η, %

No wind-wallsWith wind-walls

LITERATURE OVERVIEW

41

Unstructured mesh was used to describe the hyperbolic shape of the cooling tower’sshell (see Figure 2-29.). The cross-wind has been modelled with a uniform profile.

It was stated that at a wind velocity higher than 12.5 m/s, the separation of the air flowfrom the top edge of the shell has more influence than at the lower edge. This will helpthe buoyancy forces and increases the air flow rate through the tower. On the other hand,at lower velocities the separation at the lower edge is more dominant. In contrast, thisseparation reduces the buoyancy forces, which in turn will reduce the amount of therejected heat.

The effectiveness of the cooling tower (see Figure 2-30.) was defined as the ratio ofthe amount of heat rejected at any actual condition, to the maximum heat rejection thatoccurred at no-wind conditions (I think that the mean heat exchanger temperature waspossibly held constant for different wind speeds).

At higher wind velocities wind-walls cause separation, which will create low pressurezones at the sides of the cooling tower inlet. These zones will block the air from gettinginto the cooling tower as well as encouraging the air in the vicinity to recirculate out ofthe cooling tower. Although this behaviour does not affect the performance in a greatmanner, the shape and the location of the wind-walls need to be modified.

At a wind velocity of 15 m/s the hot air starts exiting from the rear inlet of the coolingtower. This means that the ambient air does not infiltrate sufficiently to the cooling toweras a consequence of the cross-wind (see Figure 2-31.). As the cross-wind increases, thelow pressure recirculating region will increase till the air eventually passes straight belowthe radiator without changing its direction upwards. Hereby, the thermal performance ofthe cooling tower will be affected severely.

Figure 2-31. Temperature contours of air inside the cooling tower at cross-wind velocity of 15 m/sSource: AL-WAKED ET AL. (2001)

LITERATURE OVERVIEW

42

The above overview of studies from the scientific literature may appear considerablydetailed. However, I think that it was important to shortly sum up all relevant information– parameters of the investigated cooling systems, the applied equipment and methods ofanalysis – beside pure observations and conclusions presented in different papers,because the results may depend just on these details case by case. As it is known, theeffect of wind on the performance of different cooling towers is not the same.

I have reviewed a number of studies beyond those outlined in this chapter, but due tolack of space only the most exciting ones were summarised above. Anyhow, according tomy readings I could recommend to the enquirers some further studies concerning full-scale measurements – CHRISTOPHER AND FORSTER (1969-1970), LAURAINE ET AL. (1989), LIFFICK

AND COOPER (1995), MADADNIA ET AL. (2001a), MONJOIE ET AL. (2000), RENNIE AND HAY (1990),VAUZANGES AND RIBIER (1984), WOLF (2001) – scale model tests – BENDER ET AL. (1996),BLANQUET ET AL. (1986), DASHKOV ET AL. (1996), DERKSEN ET AL. (1996), GELDENHUYS AND

KROGER (1986), MADADNIA ET AL. (1996), MADADNIA ET AL. (2001b), MADADNIA ET AL. (2001c),PEEARLMUTTER ET AL. (1996), REIZES AND MADADNIA (2001), SABATON (1984), VAUZANGES AND

RIBIER (1986), VÖLLER (1985), XU ET AL. (1989) – and numerical simulations – BERGSTROM

ET AL. (1993), CHABARD ET AL. (1993), DEMUREN AND RODI (1987), DU PREEZ AND KRÖGER (1994),DU PREEZ AND KRÖGER (1995a, b), FLASSAK (1990), GELGAND ET AL. (1990), MOUSSIOPOULOS AND

FLASSAK (1989), RAZAFINDRAKOTO AND FOURNIER (1997). Other works which also containinformation on cooling tower wind effect include BAER ET AL. (1980), BERLINER (1980),BUXMANN (1986), CHOLNOKY (1989), DU PREEZ (1992), ETZOLD AND FIEDLER (1976), FABRE AND

LEGRAND (1988), FARIVAR (1981), GOECKE (1969), HILL (2001), LINDAHL AND BUGLER (1986),MADADNIA ET AL. (2001d), MAJUMDAR ET AL. (1983, 1984), MOORE (1976, 1978), MOORE AND

GARDE (1981), OKAMOTO (1982), OKAMOTO AND SUNABASHIRI (1992), OKAMOTO AND YAGITA

(1973), RUSSELL ET AL. (1978), SLAOUTI AND GERRARD (1981), WILLA ET AL. (1980), WITTE

(1983), ZHAO ET AL. (1989). For general information on cooling tower engineering refer toVDI 2047 and BOŠNJAKOVIĆ ET AL. (1951). A collection of the formulas for calculating theheat transfer can be found in VDI-WÄRMEATLAS (1977). A study about the influence oftemperature inversion on the cooling tower thermal performance is analysed by BUXMANN

(1977).

ON-SITE MEASUREMENT OF WIND EFFECT

43

3. ON-SITE MEASUREMENT OF WIND EFFECT

The wind sensitivity of a natural draught cooling tower depends on several factors. Ahigher nominal ϑt (difference between the warm water temperature and ambient air drybulb temperature), which can be interpreted as a „driving potential” of the heat exchange,yields a lower cooling tower performance deterioration in wind. However, in case of acooling tower with higher nominal ϑt value, the effectiveness of the energy productionfor a given ambient air temperature and the investment cost of the cooling system areboth lower, so the value of ϑt has to be optimised in the design stage, wherein, amongother factors, a probability distribution of the wind speed for the given site should be alsoconsidered (high wind speeds in winter are less disadvantageous than in summer). Theeffect of wind on the performance of cooling towers is influenced also by geometricalconfigurations (vertical or horizontal heat exchanger arrangement, tower exit diameterand the corresponding air exit velocity and exit loss) and flow resistances of the heatexchangers.

In order to determine the wind effect curve on the cooling tower operation, full-scalemeasurements on a natural draught Heller type power plant dry cooling tower wereperformed. The experiment was done according to the standard guideline VDI 2049 (1981),which introduces uniform methods for the preparation, performance and evaluation ofthermal acceptance tests on mechanical and natural draught dry cooling towers. Thisguideline also provides information on test conditions, recommendations concerningclear technical requirements to be agreed upon between owner and supplier, andprimarily covers test details of the guaranteed thermal performance of a dry cooling towersubject to be confirmed by the acceptance test. Since the test conditions may vary fromthe design conditions, methods and procedures are included to allow conversion of thetest readings to establish a comparison with the guaranteed values.

The contractual Performance Guarantee Test was fulfilled successfully earlier, so thepresent survey is aimed at collecting more detailed data during a period from 15 June,2001 to 3 July, 2001. For determining the wind effect curve, the restrictions for thevalidity of measured data recommended in VDI 2049 (1981) were not applied and a newmethod was developed for evaluation.

3.1. Main data of the measured cooling system

The power plant to which the cooling system was adapted is a natural-gas-fired CCPP oftwo identical 700 MWe units. Our measurements were performed on Unit 1 coolingtower. The most important data of the cooling system are:

Location: Turkey, near Bursa city

Commissioned in: 1999

ON-SITE MEASUREMENT OF WIND EFFECT

44

Nominal heat rejected in the tower: 419.1 MW from the steam cycle + 2.3 MWdue to friction losses of cooling watercirculation (pumping power - power recoveredby the hydraulic turbines)

Nominal overall ITD0 value: 25.6 ºC

Steam temperature in the condenser: 40.6 ºC

Cooling water flow rate: 8,888.9 kg/s

Cooling tower height: 135 m

Base diameter of the tower: 121.8 m (outside diameter of deltas)

Tower throat (minimum) diameter: 67 m

Main cooling sector number: 6

Main cooling deltas: 130 pieces of 20 m long deltas + 2 pieces of15 m long deltas (above tower gate)

Main cooling delta angle: 57.8 º

Number of peak cooler cells: 12 pieces

Peak cooling deltas: 2 pieces of 5 m long deltas in each cell

Peak cooler delta angle: 40 º

3.2. Execution of the measurement

The VDI 2049 (1981) guideline clearly defines the test procedure and the physicalquantities to be measured during a Performance Guarantee Test. Table 3-1. gives anoverview about the measuring instruments and Figure 3-1. illustrates their location.

3.2.1. Calibration of the measuring instruments

Prior to the measurements, the pressure transmitters (instruments with compulsorycalibration) were calibrated one by one by the National Office of Measures in Hungarywhile the anemometer transmitters by the Budapest University of Technology andEconomics, Department of Fluid Mechanics. The thermometers were compared to eachother and to a calibrated higher precision thermometer at the site: all the thermometerswere submerged simultaneously into the water bath of a thermostat and connected to theinput channels of the Data Acquisition Device. The thermostat maintained selectedtemperature values with the help of an electric heater. The temperatures measured by thedifferent thermometers were recorded and the deviations from the exact value werecalculated. After that, regression lines were laid onto the measured points and equationswere determined for each thermometer. With these corrections the exact temperaturescould be determined from the measured values during the evaluation.

ON-SITE MEASUREMENT OF WIND EFFECT

45

3.2.2. Operation of the measuring system

The pressure measuring taps inside the condenser were provided with guide-plates whichmade it sure that the steam velocity was perpendicular to the pressure tap. The impulsetubes coming from the condenser body were connected to a collecting vessel. Themeasuring stub planned for the Performance Guarantee Test was provided with a ballvalve, therefore the absolute pressure transmitter could be connected every time withoutdisturbing the operation of the Unit.

The highly accurate resistance thermometers were connected to the Data AcquisitionDevice with four leads shielded data cables eliminating this way the effect of theresistance differences of the long measuring cables. The protecting test wells, in whichthe thermometers were placed for cooling water temperature measurement in the mainforward and return pipelines, were filled up with oil (water fill would evaporate).

The air temperature measurement may be influenced by the sunshine, by the radiationfrom the soil and from the heat exchangers. To avoid this influence, the PT 100 sensorswere inserted in double-wall cylinders. The air was forced through the cylinders bymeans of electric fans. The two thermometers held by consoles at about 2 m outwardsfrom the outer wall of the cooling tower and at about 1.5 m below the top level of the to-wer measured the ambient air dry bulb temperature in order to allow the calculation of thetemperature gradient. The lower value of the two measured temperatures was taken intoconsideration, because one of these temperatures – depending on the direction of wind –may be influenced by the hot air coming from inside of the cooling tower due to wind.

The flow sensor (type SDF-F50-1,400) had four holes at the front side and at the rearside of its specially designed profile. At the front side, in the stagnation point, the totalpressure, at the rear side the static pressure could be measured accordingly. Theirdifference gave the dynamic pressure that depends on the density of the flowing mediaand the velocity of flow according to the Bernoulli equation (it is the principle of thePitot-static tubes used for measuring the local velocity). The positions of the four holeswere designed so that the measured pressure difference (dynamic pressure) isproportional to the average velocity of the whole cross-section. This rate was taken intoconsideration by a non-dimensional transmission coefficient given by the manufacturer.

The flow rate measured by these built-in sensors was checked by flow rate valuesobtained from the characteristic curves of the main cooling water circulating pumps andthe readings of pump heads and absorbed power in the control room. In addition, the heatrejection of the tower calculated with these cooling water flow rates were compared tothe heat to be dissipated from the steam cycle according to the displayed values in thecontrol room and the characteristic curves of the steam turbine. Measurements were donealso by a portable ultrasonic flow meter at different pipe sections for determining theflow rate.

After the measurement system was assembled and checked, data were collectedcontinuously round the clock with measuring cycles of 20 s. The readings were saved into

ON-SITE MEASUREMENT OF WIND EFFECT

46

data files by the laptop computer. During the test period the sky was bright, sometimeswith smaller clouds. All cooling sectors (i.e. all main cooling delta sectors and all peakcooler sectors) and both cooling water circulating pumps were in operation. Theoperation mode of the peak cooler sectors was dry. All louvers of the tower were fullyopen and the surfaces of the heat exchangers were cleaned one week before ourmeasurements.

In order to gain information about the flow field around the tower and velocity profilesalong some individual heat exchanger panels during cross-wind conditions, a cupanemometer, a digital ducted vane anemometer and a self-manufactured Sher-disc (forstatic pressure measurement) were fixed onto a 5.5 m long rod. These instruments werecarried in hand to different locations for mensuration. During nights the rod was belayedat some places close to the tower for further data collection.

Item No. 1Measured value Dry bulb temperature of ambient air entering the cooling tower

Location In the ¼ and ¾ height of the inlet opening at the middle of each sectoraround the tower (6 × 2 = 12 pieces)

Position of measuring device Fixed by fastening belts on the horizontal rods of the cooling deltasteel structure behind the louvers, thermometer in horizontal position

Sort of measuring device PT 100 platinum resistance thermometer in double-wall shielding tubes(against thermal radiation), with electric fans

Sensor type Ahlborn P444Measuring range 0 ... 100 °CAccuracy IEC Class „A”Item No. 2

Measured value Dry bulb temperature of ambient air outside the cooling tower at133.5 m level

Location Sidewalk at the top of the tower, East side and North side (2 pieces)

Position of measuring device On console fastened to the outer wall of the sidewalk, thermometer invertical position

Sort of measuring device PT 100 platinum resistance thermometer in double-wall shielding tubes(against thermal radiation), with electric fans

Sensor type Ahlborn P444Measuring range 0 ... 100 °CAccuracy IEC Class „A”Item No. 3Measured value Temperature of cooling water entering the cooling systemLocation Machine hall

Position of measuring device Discharge pipe of cooling water pumps, near to pump outlet (2 piecesfor the two hydro machine groups)

Sort of measuring devicePT 100 platinum resistance thermometers put into the existingprotecting tubes welded into the cooling water forward lines for localthermometers

Sensor type Ahlborn P600Measuring range 0 ... 100 °CAccuracy IEC Class „A”

Table 3-1. Detailed list of measuring devices

ON-SITE MEASUREMENT OF WIND EFFECT

47

Item No. 4Measured value Temperature of cooling water returning from the cooling systemLocation Machine hall

Position of measuring device Return cooling water lines, about 1 m from machine hall wall(2 pieces for the two hydro machine groups)

Sort of measuring devicePT 100 platinum resistance thermometers put into the existingprotecting tubes welded into the cooling water return lines for localthermometers

Sensor type Ahlborn P600Measuring range 0 ... 100 °CAccuracy IEC Class „A”Item No. 5Measured value Wind velocity at 10 m level in undisturbed flowLocation At least 50 m from the cooling towerPosition of measuring device On a rod fastened to the top of a lamp postSort of measuring device Cup anemometer transmitter, (measured signal: voltage)Sensor type Thies Clima 4.3400.3Measuring range 0 ... 35 m/sAccuracy 2 %Item No. 6

Measured value Water flow rate Pressure difference of the waterflow sonde

Location Machine hall Machine hall

Position of measuring device

Built into the return lines at about4 m from the wall(2 pieces for the two hydromachine groups)

Connected parallel on the waterside to the existing differentialpressure transmitter of the plant(2 pieces for the two hydromachine groups)

Sort of measuring device

Water flow sensor operating on theprinciple of Pitot-static tubes(measured signal: pressuredifference)

Capacitive differential pressuretransmitter with power supplyunit (measured signal: current)

Sensor type Siemens SDF-F50 Gamma C719.04Measuring range 0 ... 22,500 m3/h 0 ... 250 mbarAccuracy 1 % 0.25 %Item No. 7Measured value Condenser pressureLocation Condenser upper partPosition of measuring device Connected to the measuring stubs made for the PGT

Sort of measuring device Capacitive absolute pressure transmitter with power supply unit(measured signal: current)

Sensor type Gamma C719.31Measuring range 0 ... 310 mbarAccuracy 0.25 %Item No. 8Measured value Barometric pressure of ambient airLocation Switch-gear of the cooling towerPosition of measuring device On worktopSort of measuring device Digital barometer operating on piezoresistance principleSensor type Wallace & Tiernan Diptron 3+Measuring range 0 ... 1,100 mbarAccuracy 0.04 % of full scale

Table 3-1. Detailed list of measuring devices (continued)

ON-SITE MEASUREMENT OF WIND EFFECT

48

Item No. 9Measured value Relative humidity of ambient airLocation Switch-gear of the cooling towerPosition of measuring device Hung on the outer wall of the switch-gear buildingSort of measuring device Capacitive relative humidity sensor (measured signal: current)Sensor type Ahlborn FHA 646-52Measuring range 0 ... 100 %Accuracy 2.5 %Item No. 10Function Collecting the input signals of sensorsLocation Switch-gear of the cooling tower

Position of device Connected with data cables to the sensors (4 leads to PT 100thermometers)

Sort of device Data Acquisition Device for measuring temperature, voltage andcurrent

Type Ahlborn Almemo 5590-2Resolution Less than 0.1 °C, 0.01 mAItem No. 11Function Operating the measuring system, recording the measured valuesLocation Switch-gear of the cooling towerPosition of device Connected by serial port to the Data Acquisition DeviceSort of device Laptop computerItem No. 12Function Printing the recorded dataLocation Switch-gear of the cooling towerPosition of device Connected by parallel port to the laptop computerSort of device PrinterItem No. 13Measured value Water flow rateLocation Portable

Position of measuring device Mounted on the outer wall of the pipes in which flow rate is to bemeasured

Sort of measuring device Ultrasonic flow meter using the transit-time flow measurementtechnique, with magnetic fixture

Sensor type Panametrics PT868Measuring range 0 ... 12.7 m/sAccuracy 0.5 ... 2 %Item No. 14Measured value Measurement of velocity profiles

Location Different locations inside and outside around the tower measuring theflow field

Position of measuring device On a 5.5 m rod, carried in hand

Sort of measuring device Digital ducted vane anemometer(measured signal: r.p.m.)

Cup anemometer transmitter(measured signal: voltage)

Sensor type Thies Clima 4.3405.20 Thies Clima 4.3400.3Measuring range 0.3 ... 30 m/s 0 ... 35 m/sAccuracy 2 % 2 %

Table 3-1. Detailed list of measuring devices (continued)

ON-SITE MEASUREMENT OF WIND EFFECT

49

Figure 3-1. Locations of measuring instruments

3.3. Evaluation of the thermal performance of the cooling system

For Performance Guarantee Tests on dry cooling towers the evaluation method isdescribed in the VDI 2049 (1981) guideline. In this procedure only those measurement datacan be used in the evaluation, for which the deviations from the nominal values arewithin the values stated in Table 3-2. to ensure the appropriate accuracy of thecorrections. Operational deviations during the tests shall not exceed half of thesedeviations.

Parameter DeviationInlet air temperature ±10 KWater flow rate ±10 %Thermal performance ±10 %

Table 3-2. Maximum admissible deviation of average test values from the nominal values

Furthermore, according to the guideline, test should never be carried out at fog orprecipitation of any kind and regarding corrections of the thermal performance forinversion weather conditions, Seller and Customer shall make a specific agreement.

Because up to present no sufficient experience was gained to show any lawfulnessregarding the effect of wind velocity on the thermal performance of a cooling tower, the

Natural draught drycooling tower

2

2

Motor

Pump

Hydroturbine

Boiler feedwater

Jet condenser

Steam turbine

Switch-gear building

5

3

3

4

46

6

7

910

11

8

10 m

1

12

ON-SITE MEASUREMENT OF WIND EFFECT

50

VDI 2049 (1981) guideline recommends that the mean wind velocity at cooling tower topelevation should not exceed 3 m/s and the instantaneous wind velocity should not exceed6 m/s more than 20 times during the test period of one hour used for performanceevaluation.

Considering the fulfilment of the above listed conditions a one-hour period should beselected from the collected data for evaluation and the mean values of the data during thisperiod are considered as the result of the measurement.

Since the measured conditions usually deviate from the nominal values, the measuredheat dissipation should be transformed before comparing it with the guaranteed thermalperformance. The thermodynamic behaviour of the cooling plant at operating conditionsdeviating from the design point is shown in the form of diagrams – a basic diagram andcorresponding correction curves – considering the most important variable parameters.These figures had already been supplied by EGI - Contracting Engineering Co. Ltd.before the Performance Guarantee Test measurement in 1999 as part of the coolingsystem purchase contract.

The evaluation procedure is outlined below:

Calculation of the measured overall ITDm from the condenser temperature, Ts

(determined from the measured absolute pressure of the condenser, using tables forsaturated steam properties), and ambient air temperature, Ta:

as TT −=mITD (3.1)

If the air inlet dry bulb temperature deviates from the nominal ambient airtemperature, Tan, determining the correction:

( )a

aana TQTTQ

∂∂

⋅−=Δ (3.2)

The correction curve to be used shows the relationship ( )mITD;wa

mfTQ

=∂∂ for the

measured cooling water mass flow rate, mw, and overall ITDm in case of nominalheat rejection and nominal barometric pressure, pbn.

If the barometric pressure, pb, deviates from the nominal value, pbn, determining thecorrection:

( )b

bbnb pQppQ

∂∂

⋅−=Δ (3.3)

The second correction curve shows the relationship ( )mITD;wb

mfpQ

=∂∂ for the

measured cooling water mass flow rate and overall ITDm in case of nominal heatrejection and nominal ambient air temperature, Tan.

ON-SITE MEASUREMENT OF WIND EFFECT

51

With minor deviations from the design point or minor influence of the main

parameters mw and ITDm on the corrections, the functions aT

Q∂∂ and

bpQ

∂∂ can be

approached by constants.

Calculation of the measured heat dissipation, Qm:

( )w2w1wwm TTcmQ −⋅⋅= , (3.4)

where cw is the specific heat of water at the average cooling water temperature, Tw1

is the temperature of cooling water entering the tower and Tw2 is the watertemperature returning from the tower.

Calculation of the corrected heat dissipation duty:

bam QQQQ Δ−Δ−=∗ (3.5)

Using this Q* value and the relative water mass flow, mw/mwn, a target ITDtarget

value is ascertained from the basic diagram, which shows the relationship( )targetITD;wmfQ =∗ related to design ambient air dry bulb temperature, Tan, and

barometric pressure, pbn.

Guarantee is fulfilled if the following condition has been complied with:

targetm ITDITD

*m QQ

≥ (3.6)

For measurement evaluation an advanced computer program was developed. It made itpossible to evaluate posteriorly within a relatively short time also the previousPerformance Guarantee Test from 1999. The total tolerance of the measurement due tothe accuracy of the instruments was determined to be ±2.3 %. Detailed description ofcalculating the total tolerance of an acceptance test is provided in VDI 2049 (1981).

3.3.1. Another evaluation method

BUXMANN AND HAMANN (1974) described another method to calculate the diagrams of forceddraught and natural draught cooling towers. It establishes the thermal efficiencycoefficient deviating from the design point and finally the desired parameters such asthermal performance and overall ITD. This allows conversion from design point to anyother operating point. This function contains only values for pressure, temperature andviscosity of air as well as water temperatures and water flow rates to be determined frommeasurements.

3.4. Wind effect curve obtained from full-scale measurements

In our calculation aimed at determining the effect of wind on the cooling towerperformance, the limitations recommended in VDI 2049 (1981) for the validity of the data

ON-SITE MEASUREMENT OF WIND EFFECT

52

to be used for evaluation were not applied, as our computing method was different fromthat of the guideline.

The effect of wind is characterised by the deterioration of tower ITD with constantheat rejection due to wind. The tower ITD is defined as

aw1 TT −=ϑt , (3.7)

where Tw1 is the temperature of cooling water entering the tower and Ta is the ambient airdry bulb temperature.

The database of the measurement evaluated in the frame of this survey containedalmost 60,000 data series altogether. The determination of the wind effect curve wasmade in the following manner:

1. Several one-hour long measuring periods were selected for evaluation,

2. The arithmetic mean values of measured data during these one-hour periods werecalculated,

3. The heat rejection and the tower ITD were determined from the average values foreach one-hour periods according to equations (3.4) and (3.7), respectively.

4. For every one-hour measuring period corrections were calculated (instead of usingcorrection curves like in the VDI 2049 (1981) evaluation method) for all measuredfactors influencing the operation of the cooling tower except wind velocity (i.e. thetemperature and relative humidity of ambient air, barometric pressure, coolingwater mass flow rate, heat rejection and possibly the inversion in the atmosphere) inorder to transform the measured tower ITD, ϑt, to nominal operating conditions.The corrections were based on cooling tower sizing calculations, which rely ontheoretical relationships, earlier laboratory and full-scale measurements and thehalf-century experiences of EGI - Contracting Engineering Co. Ltd. in power plantnatural draught dry cooling towers.

5. The differences between the average tower ITD values obtained in the previous step– which relate to nominal ambient air temperature, barometric pressure, coolingwater mass flow rate and heat rejection – and the nominal tower ITD were relatedone by one to the nominal tower ITD. These Δϑt/ϑt0 values were plotted against theaverage wind velocities in the relevant one-hour periods.

The effect of wind obtained by this method is shown in Figure 3-2. (curvef (v10)measured). Here ϑt0 is the tower ITD at nominal and wind-off conditions, while Δϑt isthe one-hour average value of the change in tower ITD at a wind speed v10 measured at10 m level above ground and averaged also over one hour. In this figure, the tower isconsidered to operate at nominal conditions in wind. Beside determining the wind effectcurve the main purpose of this calculation was to establish a figure that can be used forcomparison with results obtained by Computational Fluid Dynamics modelling.Therefore, curve f (v10)measured in Figure 3-2. which relates to average wind speeds during

ON-SITE MEASUREMENT OF WIND EFFECT

53

one hour was transformed to curve f (v10)theoretical which relates to constant wind speeds.The relationship between the two curves is the following:

( ) ( ) ( )∫∞

⋅⋅=0

dΦ 1010ltheoretica10measured10 vvvfvf (3.8)

It is obvious, that a wind effect curve which is calculated in theory for constant windspeeds, f (v10)theoretical, cannot be applied to real conditions, because the wind blows withvariable speeds. The wind effect strongly depends on the meteorological features of thewind, like velocity profile with height or statistical distribution.

If, for example, a steady wind blew with a velocity of 3 m/s, the wind effect would becc. 0.4 %. If, on the contrary, this average 3 m/s wind speed arose only from windvelocities of 0 and 6 m/s blowing for equal time periods, the resulting wind effect wouldlie on the dashed line at a value of cc. 1.2 %. The wind effect with a real statistical windvelocity distribution would result somewhere between these two values, that is betweenthe continuous line obtained with constant wind speeds and the dashed line (as the windeffect curve is convex from beneath, i.e. wind velocities greater than 3 m/s cause moredeteriorating effect than wind velocities lower by the same extent than 3 m/s have smallerwind effect). In addition, according to the differences between the dashed-dotted and thecontinuous curves in Figure 3-2., it can be read out that at 3-3.5 m/s average wind speeds,for example, the wind-gusts increase the wind effect by approximately 50 %.

0

0.02

0.04

0.06

0.08

0 1 2 3 4 5 6 7 8w� [m/s]

Figure 3-2. Wind effect curve obtained from measurement

The theoretical wind effect curve is not obtainable from the measurements directly, butwith calculations in reversed direction it can be derived from the average values,f (v10)measured, and the density function, Φ(v10). The equation for the f (v10)measured windeffect curve was obtained by fitting a power function trendline on the data points by theleast squares method. In order to determine the density function of wind velocity, thedistribution function of instantaneous wind velocities was composed and a sixth orderpolynomial trendline was created for it by the least squares method, the equation of whichwas derived. Figures 3-3. and 3-4. illustrate the distribution and density functions of themeasured instantaneous wind velocities, respectively.

f (v10)theoretical

f (v10)measured

Δϑt/ϑ

t0

v10, m/s

ON-SITE MEASUREMENT OF WIND EFFECT

54

0

1

2

3

4

5

6

7

0 0.2 0.4 0.6 0.8 1Probability [-]

v w [m

/s]

Figure 3-3. Distribution function of the measured instantaneous wind velocities

0

0.10.2

0.3

0.4

0.50.6

0.7

0 1 2 3 4 5 6

vw [m/s]

Prob

abili

ty d

ensit

y

Figure 3-4. Density function of the measured instantaneous wind velocities

A detailed description of this analysis is summarised in CSABA AND KAPÁS (2002).

Probability

v 10,

m/s

Prob

abili

ty d

ensi

ty, s

/m vaverage = 2.54 m/s

v10, m/s

COMPUTATIONAL FLUID DYNAMICS MODELLING

55

4. COMPUTATIONAL FLUID DYNAMICS MODELLING

In order to investigate different conceptions aimed at improving cooling tower behaviourin wind, first the Computational Fluid Dynamics (CFD) model of the aforementionedBursa cooling tower was built, with which the effect of wind on the tower could bedetermined by calculations for some selected stationary wind speeds. The wind effectcurve obtained this way was compared to the theoretical wind effect curve for constantwind speeds resulting from full-scale measurements (Figure 3-2.). Below we will see thatthe agreement of these two curves is more than satisfactory, so the subsequent wind effectanalyses can be based on the CFD model.

In this chapter this cooling tower CFD model is described in detail and the results ofthe calculations are also presented. The applied CFD software was FLUENT 6.0.20.

4.1. Geometry

As the first step, geometry was created in 3D based on data used in sizing calculations(KESZTHELYI, 1996), design drawings of the cooling system and data provided by differentsuppliers of equipment (see Figure 4-1.). In CFD modelling, before starting the geometrybuilding, decisions should be made what details are important and will be included andwhat will be neglected. In addition, the structure of the whole model should be imaginedin advance, because the possibilities in the subsequent steps depend on the compositionof different geometrical shapes (meshability of faces and volumes, application ofboundary conditions). Efforts were made to take into account the main features of the realconstruction, however, at some points the exact geometrical shapes had to be simplified:

The cooling delta frame structure was not included in the model, but the location ofthe heat exchangers corresponds to the real arrangement (the heat exchangercolumns were considered as so many individual fluid zones as their number in onehalf of the real cooling tower).

The louvers in front of the heat exchangers and the tower X-legs were modelled byporous jump boundary conditions across simple planes.

The effect of cooling water piping was modelled by porous zone boundarycondition with appropriate pressure loss coefficient in the air flow.

The outlet of the peak cooler cells was created with rectangular cross-section in themodel without the transition piece to circular cross-section. The fans were modelledby the fan model of FLUENT at the exit planes.

The cooling tower shell was considered as a wall with „zero” thickness, and itsprofile was formed by a spline interpolated through the average of outer and innershell diameters at the bottom, throat and exit.

The power plant buildings around the tower were not incorporated in the model.

COMPUTATIONAL FLUID DYNAMICS MODELLING

56

Figure 4-1. Cooling tower geometry

SymmetrySymmetry

Velocityinlet

Symmetry

Outflow

Wall

2200m

1500m

1400m

Figure 4-2. The computational domain

The model consists of 369 individual volumes and contains 2,627 faces. Thesevolumes and faces were grouped into 183 cell zones and 1,488 face zones, respectively,so the number of boundary conditions was somewhat reduced.

Symmetry

Symmetry

Velocity inlet

Outflow

2,200 m

1,500 m

1,400 m

Symmetry

Wall

COMPUTATIONAL FLUID DYNAMICS MODELLING

57

Assuming symmetrical thermal and flow fields in the model, only one half of thecooling tower was modelled with a symmetry boundary condition. The dimensions of thecomputational domain were determined according to the recommendations by STRAW

(2001): if H denotes the greatest size of the cooling tower, the domain should exceed 5·Hupstream, 10·H downstream and 6·H upwards the tower as usual. The domain isillustrated in Figure 4-2., where the types of the boundary conditions applied at the pe-ripheries are also indicated and the model is mirrored against the central symmetry plane.

4.2. Mesh

During mesh generation, much attention was paid to comply with mesh qualityrequirements recommended in FLUENT User’s Guides (FLUENT, 2001a-e). In order tohave an appropriate resolution of the flow field in the vicinity of and inside the coolingtower, the computational domain was discretised into a large number of finite volumecells. Detail from the mesh is illustrated in Figure 4-3.

Figure 4-3. Bottom region of the cooling tower – quadrilateral face mesh elements

Figure 4-4. Finite volume cells coloured by their Equiangle Skew quality metric

COMPUTATIONAL FLUID DYNAMICS MODELLING

58

The faces were meshed with quadrilateral mesh elements using Map, Pave and Tri Pri-mitive meshing schemes, and the volumes were meshed with hexahedral elements usingMap and Cooper schemes. The model contains altogether 752,549 finite volume cells.

For every heat exchanger zone two cells were generated in the direction of air flow inorder to be able to take into account the effect of two passes of cooling water flow in thetube bundles (see Figure 4-4.).

Figure 4-5. Location of interface boundary conditions

To avoid extremely large number of cells causing long CPU time for obtaining aconverged solution, the reduction of the number of cells in the model was facilitated inthree ways:

By applying interface boundary conditions, which means that two identical facesexist at the same location, hereby different mesh sizes can be used on the two sidesof the respective plane while the faces are treated as simple internal faces withoutany effect to the air flow. Interfaces were used at two places in our model (seeFigure 4-5.) and in both cases the finer mesh was coming from the proximity of thecooling tower and the coarser mesh was extending into the bulk of thecomputational domain. In this manner, the very small cells generated for thedetailed tower geometry were not retained through the domain. (Other internalfaces are shared commonly by their two adjacent volumes so the mesh size must bethe same at the two sides of the faces.)

The inner and outer surfaces of the walls inside the model have identical shapes,but are disconnected and consist of different faces, so the mesh sizes on the two

Interface 1: annulus shaped surfaceoutside the tower, cc. 1 m above thetop of heat exchangers

Interface 2: polygon surface,50 m above the tower outlet

COMPUTATIONAL FLUID DYNAMICS MODELLING

59

sides of the walls can be different. Nevertheless, every wall was modelled with„zero” wall thickness.

Using size functions or grading mesh nodes on the edges (i.e. allocatingcontinuously increasing edge mesh intervals) in order to increase gradually the cellsize for regions which are far from the tower.

4.3. Model settings

In this section the applied model settings are briefly introduced. Readers who are notfamiliar in FLUENT can find detailed description of the definitions for specific concepts,theoretical background and model equations, calculation method for different boundaryconditions, solution techniques as well as references to the concerning scientific literaturein FLUENT (2001a-e).

4.3.1. Grid manipulation

The domain was reordered using reverse Cuthill-McKee method. Hereby a bandwidthreduction of 53.13 was achieved with maximum cell distance of 3,977. However, aftercreating the interface zones the maximum cell distance increased to 24,039.

4.3.2. Turbulence modelling

The calculations were done with the segregated implicit solver assuming steady flow in3D. The turbulent nature of flow was taken into account by the RNG k-ε turbulencemodel with non-equilibrium wall functions and enabled differential viscosity model, swirldominated flow and full buoyancy effects options.

4.3.3. Material properties

During the simulations the material properties of air were computed according to thefollowing considerations:

For calculating the density the equation of state for incompressible ideal gases wasused (the air density is a function of the temperature only),

The specific heat was defined as a piecewise linear function of temperature,

The thermal conductivity was modelled using the kinetic theory,

The viscosity was calculated by the three coefficient Sutherland law,

The molecular weight was constant, 28.966.

Properties for solid materials were set as constants taken from RAŽNJEVIĆ (1964).

4.3.4. Operating conditions

The operating pressure was given as 99,277.2 Pa at a point 100 m upstream from thecentreline of the cooling tower at ground level. The gravitational acceleration was

COMPUTATIONAL FLUID DYNAMICS MODELLING

60

specified as 9.81 m/s2. For the Boussinesq approach an operating temperature of288.16 K and an operating density of 1.2 kg/m3 was entered.

4.3.5. Wall boundary conditions

At walls zero heat flux boundary condition was applied (adiabatic walls). For momentumequation no slip shear condition was prescribed and a wall roughness height wasspecified. In FLUENT an „equivalent” sand-grain roughness height should be used withthe default roughness constant of 0.5. When determining the „equivalent” sand-grainroughness height for the physical roughness height of different walls, recommendationsin SCHLICHTING (1965) were studied.

4.3.6. Thermal modelling of heat exchangers

It is impractical to model individual fins and tubes of a heat exchanger core. In principle,heat exchanger cores add heat and introduce a pressure drop to an air stream. InFLUENT, lumped parameter models are used to account for these effects with thefollowing assumptions:

The heat exchanger effectiveness, ε HE, is defined for a complete heat exchanger,and can be applied to a small portion of the heat exchanger represented by acomputational cell,

The air capacity rate, (m·cp)air, is less than the coolant capacity rate,

The cell temperature, Tcell, (i.e. the cell centroid value) can be used instead of thetemperature of the fluid entering the cell,

Flow acceleration effects are neglected in calculating the pressure loss coefficient,

The coolant temperature must be higher than the air temperature,

The coolant is restricted to a single phase.

The heat transfer is modelled by a heat source in the energy equation and the pressureloss is modelled by a momentum sink in the momentum equation, respectively.

In a typical heat exchanger core, the coolant temperature is stratified in the direction ofcoolant flow. As a result, heat rejection is not constant over the entire core. In FLUENT,the fluid zone representing the heat exchanger core is subdivided into macroscopic cellsor „macros” along the coolant path (see Figure 4-6.). The coolant inlet temperature toeach macro is computed and then subsequently used to compute the heat rejection fromeach macro. This approach provides a realistic heat rejection distribution over the heatexchanger core.

The fluid zones representing the heat exchanger cores were sized to the dimension ofthe core itself in our cooling tower model. The dimensions of the heat exchanger core arethe height (defined along the direction of the coolant inlet), width (defined along the pass-to-pass direction) and depth.

COMPUTATIONAL FLUID DYNAMICS MODELLING

61

As part of the setup procedure, the coolant path was defined separately for each heatexchanger zones to ensure the cross-counter flow of the air and cooling water (132 maincooling and 24 peak cooling heat exchangers). To define the coolant direction and flowpath, direction vectors were specified for the coolant inlet direction and the pass-to-passdirection. The number of macros per pass was specified as 6 and 5 in case of maincooling and peak cooling heat exchanger zones, respectively. Based on this and thespecified number of passes, 2 for each heat exchanger zone, and the correspondingcoolant inlet and pass-to-pass directions, the macros were constructed automatically. Thephysical properties and operating conditions of the core (heat exchanger effectiveness,coolant flow rate) were defined separately for main coolers and peak coolers.

Mac

ro 4

M

acro

5

Mac

ro 6

M

acro

7

Mac

ro 3

M

acro

2

Mac

ro 1

M

acro

0

Coolant path

Figure 4-6. Heat exchanger core discretised into 4 × 2 macros (example)

The model computes the total heat rejection for a fixed coolant inlet temperature ineach heat exchanger zone in the following way. Heat rejection is computed for each cellwithin a macro and added as a source term to the energy equation for the air flow. Theheat transfer for a given cell is computed from

( ) ( )cellinairpcell TTcmq −⋅⋅⋅= HEε , (4.1)

where ε HE is the heat exchanger effectiveness,

(m·cp)air is the air capacity rate (flow rate × specific heat),

Tin is coolant inlet temperature of macro containing the cell,

Tcell is cell temperature.

The heat rejection from a macro is calculated by summing the heat transfer of all thecells contained within the macro:

∑=

macroincellsall

cellmacro qq (4.2)

COMPUTATIONAL FLUID DYNAMICS MODELLING

62

The total heat rejection from the heat exchanger core is computed as the sum of theheat rejection from all the macros:

∑=macrosall

macrototal qq (4.3)

The coolant inlet temperature to each macro (Tin in equation (4.1)) is computed basedon the energy balance of the coolant flow. For a given macro

( ) ( )inoutcoolantpmacro TTcmq −⋅⋅= , (4.4)

where Tin and Tout are the inlet and outlet temperatures of the coolant in the macro,respectively. The value of Tout then becomes the inlet temperature to the next macro. Thefirst macro (macro 0) is assumed to be where the coolant enters the heat exchanger core.The coolant inlet temperature for macro 0 is a given value.

Since the heat exchangers connected in parallel represent significant flow resistancesin the cooling water circuit, the flow rate can be considered equally distributed amongthem. However, the coolant mass flow rate is adjusted to be different for the main coolingand peak cooling heat exchangers. According to the design conditions 32.6 kg/s and4.797 kg/s was specified for the water flow rate in the main coolers and peak coolers,respectively. The specific heat and inlet temperature of water were set as cp =4,186.8 J/(kg·K) and Tin0 ( = Tw1) = 312.68 K for each heat exchanger zone.

The effectiveness of the heat exchanger core (ηHE in equation (4.1)) was defined as apiecewise-linear function of air velocity through the heat exchanger zone. Air velocitiesof 0.1, 1, 2, 3, ..., 15 m/s were used to specify this function at 16 points. The air-side heattransfer coefficient, αair, was available from earlier laboratory tests in the form of:

B1air LAα ⋅= (4.5)

here A and B are constants determined from laboratory tests,

L1 is the specific air mass flow rate to the tested heat exchanger.

In equation (4.5) αair is related to the frontal area of the heat exchanger panel and it ischaracteristic for a heat exchanger with 6 tube rows.

Since we had two cells in the pass-to-pass direction of the heat exchanger zones (seeFigure 4-4.) the formula for calculating the heat exchanger effectiveness, ε HE, wasdeduced with the assumption of two tube rows (i.e. one tube in one pass):

airWk

−

−= e1εHE (4.6)

where

( )HE

airpair A

cmW

⋅= in [W/m2, K] units, (4.7)

AHE [m2] is the frontal area of the heat exchanger,

COMPUTATIONAL FLUID DYNAMICS MODELLING

63

waterair α1

α1

121

+⋅=k (4.8)

The water-side heat transfer coefficient, αwater, is related to the frontal area of the heatexchanger panel and can be considered as constant since the water flow rate is alsoconstant.

As the main cooler and peak cooler heat exchangers have different heat transfercharacteristics (different A and B constants in equation (4.5)) their effectiveness wasdefined by different values.

4.3.7. Pressure drop in heat exchanger zones

The heat exchanger zones were considered as porous zones and the pressure drop acrossthem was defined by the power law model:

vvvC

spS iC

i ⋅⋅−=ΔΔ

= 1

0 (4.9)

where Si is the source term for the ith (x, y or z) momentum equation,

Δp is the pressure drop along path Δs travelled by the air parcel,

C0 and C1 are user-defined empirical coefficients,

vi is the ith component of the velocity vector.

In the power-law model the pressure drop is isotropic. Since the flow direction can beconsidered to be perpendicular to the frontal plane of heat exchangers (i.e. the air is tryingto cover the shortest distance across the porous zone), and if Δp denotes the pressuredifference before and after a cooling element, then Δs is equal to the depth of the heatexchanger core. The coefficients C0 and C1 were determined from the relationship forcalculating the pressure drop of heat exchangers obtained as result of earlier laboratorymeasurements and expressed in form:

Q1LPΔp ⋅= (4.10)

here P and Q are constants determined from the laboratory tests,

L1 is the specific air mass flow rate to the tested heat exchanger.

Since the main cooler and peak cooler heat exchangers are of different types, thepressure drop was calculated by different constants.

Porosity of these zones was specified as 1.

4.3.8. Turbulence modelling in the heat exchanger zones

The effect of heat exchanger fins and tubes on turbulence was taken into consideration byapplying User Defined Functions (UDFs) for calculating „fixed values” for the turbulence

COMPUTATIONAL FLUID DYNAMICS MODELLING

64

kinetic energy and dissipation rate in these zones. The assumptions according to whichthe formulas were derived are summarised below.

The isotropic turbulence kinetic energy, k, by definition is

( ) 223

22221

um

wvumk ′⋅≈

′+′+′⋅⋅= (4.11)

where m is the mass of a fluid parcel,

u', v', w' are the velocity fluctuations in the x, y, z directions, respectively.

FLUENT (2001e) suggests that the turbulence intensity, I, which is defined as the ratio ofthe root-mean-square of the velocity fluctuations, u', to the mean flow velocity, uavg, canbe estimated at the core of a fully-developed duct flow according to the followingformula derived from an empirical correlation for pipe flows:

8/1Re16.0 −⋅=′

≡HD

avguuI (4.12)

The Reynolds number, HDRe , is based on the hydraulic diameter, DH, of a rectangular

channel bounded by two adjacent cooler tubes and two adjacent plate fins in the heatexchanger core:

ν

⋅= Havg

D

DuH

Re (4.13)

DH was determined and incorporated into the formulas according to the heat exchangergeometry. The direction of the air flow was supposed to be perpendicular to the frontalplane of heat exchangers, therefore, uavg, can be calculated from the u, v and w velocitycomponents:

222 wvuuavg ++= (4.14)

The kinematic viscosity, ν, is calculated according to the Sutherland law appliedthroughout in the model for the dynamic viscosity of air:

STST

TT

++

⋅⎟⎟⎠

⎞⎜⎜⎝

⎛⋅

ρμ

=ν 0

2/3

0

0 , (4.15)

where ρ [kg/m3] is the density, T [K] is the static temperature, μ0 [kg/m,s] is a referencevalue, T0 [K] is a reference temperature and S [K] is an effective temperature, called theSutherland constant.

With equations (4.8) to (4.12) the turbulence kinetic energy can be calculated for everyindividual cell in the heat exchanger zones. The variables u, v, w, ρ and T are supplied bythe solver at the actual iteration for each cells in the heat exchanger zones.

The turbulence dissipation rate, ε, is determined according to the relationship used inFLUENT:

COMPUTATIONAL FLUID DYNAMICS MODELLING

65

l

2/34/3 kC ⋅=ε μ , (4.16)

where Cμ is an empirical constant specified in the turbulence model. To determine theturbulence length scale, l, an approximate relationship suggested by FLUENT (2001e) wasused:

HD⋅= 07.0l (4.17)

Since the main cooler and peak cooler heat exchangers are of different types, the fixedvalues for turbulence quantities were calculated by different constants.

4.3.9. Wind profile

The most important boundary condition at the periphery of the domain was the velocityinlet plane, because here profiles of fully-developed atmospheric flows had to beprescribed.

We can find several classifications and recommendations relating to profiles ofatmospheric flows in the literature (KAPÁS, 2002). However, when applying them in aFLUENT model as inlet conditions, it was observed that they change when going awayfrom the inlet plane as long as a fully-developed flow was achieved. This was notacceptable in our survey, because the flow pattern approaching the cooling tower wasdifferent in this case from that prescribed at the inlet plane of the domain. To solve thisproblem, first we generated fully-developed flow profiles for different ambient airvelocities by FLUENT using a simple periodic model, which is illustrated in Figure 4-7.After that the flow variables (velocity, turbulence kinetic energy and turbulencedissipation rate) were exported to a simple text file which was read in the cooling towermodel and linked to the velocity inlet plane boundary conditions. The air temperature wasspecified to be constant along height, 288.16 K.

Periodic boundaries

Symmetry

Symmetry

Symmetry

Wall

1500m

25m1400m

Figure 4-7. Auxiliary model used to generate fully-developed flow profiles

COMPUTATIONAL FLUID DYNAMICS MODELLING

66

The periodic model shown in Figure 4-7. puts back the flow variables obtained at theoutlet plane onto the inlet plane. The iteration proceeds until the changes in thesuccessive steps are negligible. The depth of this computational model is equal to onefinite volume cell in the wind direction, but its inlet plane has the same extensions as ourlarge model with the cooling tower in Figure 4-2. Thus, when applying the profiles of theflow variables obtained by this method to the inlet plane of the model in Figure 4-2., theflow pattern prescribed at the inlet is not changing along the wind direction till the effectsof the cooling tower structure appear.

The fully-developed shape of the wind profile mostly depended on the roughness ofthe ground and to a lesser extent the wind velocity had also an effect. We generated windvelocity profiles for two ground roughness heights: 0.05 m and 0.4 m (sand-grainroughness height). When these profiles were compared to the power law formula of thewind speed,

n

refref zz

vv

⎟⎟⎠

⎞⎜⎜⎝

⎛= , (4.18)

for the exponent n 0.087 (0.05 m roughness height) and 0.102 (0.4 m roughness height)were obtained by fitting a trendline on the profiles using the least squares method.

4.3.10. Fan boundary condition

The peak cooler cells were simulated in dry operation mode, so in the fan boundarycondition a pressure jump of 0 Pa was specified.

4.3.11. Air tunnel of peak coolers

The air tunnel through which air flows from the bottom of main cooling deltas into thepeak cooler cells was treated as a porous zone in the FLUENT model. In the momentumequation the pressure loss due to the main cooling sector aboveground ring piping and thesteel structures (heat exchanger and pipe supports, stiffening cross rods in the peak coolercells) in this air tunnel was taken into account by a source term having form of the powerlaw model:

vvvC

spS i

i ⋅⋅−=ΔΔ

= 2 (4.19)

The value of the constant C was determined so that for nominal flow rate to the peakcoolers the same pressure drop occurs through the tunnel as that considered in the sizingcalculation of the tower.

4.3.12. Tower X-legs

The tower X-legs were modelled by a porous face through which the pressure loss wascomputed according to the following equation:

COMPUTATIONAL FLUID DYNAMICS MODELLING

67

mvCp Δ⋅⋅ρ⋅⋅=Δ 22 2

1 , (4.20)

where C2 [m-1] is the pressure-jump coefficient, v [m/s] is the velocity normal to theporous face and Δm [m] is the thickness of the medium.

The pressure-jump coefficient, C2, was determined using the drag coefficient of asquare-shaped body, CD, from GRUBER AND BLAHÓ (1981). C2 was calculated from CD withthe assumption that the same force is acting upon the X-legs and the porous face:

mAA

CCfaceporous

legsXD Δ⋅⋅= −

2 , (4.21)

where AX-legs [m2] is the frontal area of the X-legs in one half of the cooling tower,Aporous face [m2] is the area of the porous face in the CFD model.

4.3.13. Louvers

The pressure loss through the louvers was also modelled by a porous face. The thicknessof this face was set to 0.06 m and the pressure-jump coefficient, C2, was determined as16.33 m-1 according to laboratory tests of the louvers in case of full opening.

4.3.14. Solver settings

The discretisation schemes for the solved equations are summarised in Table 4-1. Thefirst order upwind scheme resulted in better convergence and stability than the secondorder upwind scheme.

Variable SchemePressure PRESTO!Pressure-Velocity Coupling SIMPLEMomentum First Order UpwindTurbulence Kinetic Energy First Order UpwindTurbulence Dissipation Rate First Order UpwindEnergy First Order Upwind

Table 4-1. Discretisation schemes

4.4. Solution initialisation and monitoring

Solution was initialised by means of an interpreted UDF written in the C programminglanguage. In this procedure flow variables of a fully-developed flow (velocitycomponents, turbulence kinetic energy and turbulence dissipation rate) and a constant airtemperature of 288.16 K were set to all finite volume cells in the domain. After that anupward velocity of 5.81 m/s and a temperature of 306.3 K (corresponding to the nominaloperating conditions of the tower) were specified manually in the cells inside the coolingtower. With this it was ensured that a reasonable initial guess was provided for theiteration.

COMPUTATIONAL FLUID DYNAMICS MODELLING

68

In order to judge whether convergence is achieved, the scaled residuals of the transportequations and the air mass flow rate at the tower exit plane were plotted during theiterations. As the solution converges, the residuals decay to some small value and thenstop changing. Similarly, the tower air mass flow rate reaches its final value.

Usually there was a little instability even after 2000 iterations. In these cases thecalculations were stopped and evaluated after several oscillation periods when the airmass flow rate was close to the average value determined for the last few periods (seeFigure 4-8.). The oscillations of the tower air mass flow rate caused cc. 7 MW (1.66 % ofthe nominal heat rejection) variations in tower heat rejection.

450055006500750085009500

1050011500

0 500 1000 1500 2000 2500 3000

Iteration

Mas

s Flo

w R

ate

[kg/

s]

wind speed: 16 m/s

wind speed: 9 m/saverage

Figure 4-8. Convergence history of mass flow rate at tower outlet

At some wind velocities (e.g. 9 m/s) the oscillations were very small. However, at awind speed of 1.5 m/s at 10 m above ground a very unstable tower air mass flow rate wasexperienced (see Figure 4-9.). It was due to the periodic entrainment of ambient cold airinto the tower at the outlet cross-section. This is a time-dependent phenomenon, and thiswind velocity was not considered in determining the wind effect curve.

9000

9500

10000

10500

11000

11500

12000

0 500 1000 1500 2000 2500 3000

Iteration

Mas

s Flo

w R

ate

[kg/

s]

Figure 4-9. Convergence history of mass flow rate at tower outlet (wind speed: 1.5 m/s)

COMPUTATIONAL FLUID DYNAMICS MODELLING

69

4.5. Results

In this section we will see that measurement data presented in Chapter 3 are wellapproximated by the CFD analysis. Furthermore, all characteristics experienced and/orassumed previously, together with some new observations concerning fluid flow througha natural draught dry cooling tower in wind can be found in the numerical reports,solution animations and visualised flow fields of the CFD results.

4.5.1. Flow visualisation

The flow field can be visualised by contours, velocity vectors and path lines. Carefulanalysis of these distributions at different cross-sections of the domain proved that thecalculated flow variables are reasonable.

In Figure 4-10. we can see the velocity distribution in the x = 0 m plane, which isperpendicular to the wind direction and passes through the centreline of the tower. At thebottom, a hemisphere-shaped space can be observed in the middle of the tower, in whichvery low velocities are prevailing. When the air flow begins to turn upwards after passingthrough the heat exchangers, a separation bubble is arising due to the abrupt transition ofthe annulus-shaped tower skirt (covering) into the hyperbolic shell. The hot air is risingupwards as a free jet above the tower exit plane. In Figure 4-10., β is the angle of thethermal contraction. If β is greater, then the cooling tower plume contracts into a smallercross-section due to the intensive upward acceleration of warm air. At the same time, inan inverse situation, β characterises the aptitude of the cold air to sink down in warmersurrounding air (see also Figure 4-18.).

Figure 4-10. Contours of velocity magnitude [m/s] in the x = 0 m plane, wind speed: 0.3 m/s

Figure 4-11. shows the temperature distribution of the plume in the symmetry plane ofthe domain for 6 m/s wind speed (at 10 m above ground in undisturbed flow). The

β

Separation bubble

Tower skirt

COMPUTATIONAL FLUID DYNAMICS MODELLING

70

temperature decrease due to the mixing of the free jet with the ambient air is wellillustrated. (In this figure, the air temperatures greater than 291 K are coloured reduniformly in order to obtain better visibility of the plume.)

Figure 4-11. Contours of static temperature [K] in the y = 0 m plane, wind speed: 6 m/s

Figure 4-12. displays the static pressure contours (overpressures) in a horizontal plane12 m above ground (it is approximately at the middle of the height of the heatexchangers) for 6 m/s wind speed.

Figure 4-12. Contours of static pressure [Pa] in the z = 12 m plane, wind speed: 6 m/s

The highest overpressure is prevailing at the frontal stagnation point. Its value is cc.9 Pa, which is slightly less than the half of the dynamic pressure of the wind. In theregion of cc. the 19th cooling column (α = 26º) the pressure is equal to the ambient static

Wind α

Wind

COMPUTATIONAL FLUID DYNAMICS MODELLING

71

pressure (i.e. the overpressure is zero). The location of the region with the lowest pressureis nearly perpendicular to the direction of wind. Here the value of depression is cc. -49Pa.

In the case of a solid cylinder, one can expect in viscous flows a separation of theflowing fluid from the wall of the cylinder at a certain point, and at the point opposite tothe frontal stagnation point the velocity is not zero, so the static pressure is lower than thetotal pressure. However, measurements on the operating cooling towers show thatalthough the Reynolds number is large (if the characteristic length is taken as the towerdiameter, the Reynolds number is ~ 4.8·107), the flow does not separate from the surfaceat the level of heat exchangers, and a higher pressure region develops also at the rear sidedownwind the tower, but not directly on the tower circumferential surface. It can beassumed that separation is impeded by the continuous sucking of the air flowing in aboundary layer directly near by the heat exchangers into the tower. The pressuredistribution with high pressure regions at the windward as well as at the leeward side canbe observed also in Figure 4-12.

Above the heat exchanger modules, where this suction of air into the tower is notpresent, the flow separates and the high pressure region does not evolve at the rear side ofthe tower. At elevations above the top of heat exchangers (z > zHE,top) the flow pattern issimilar to that of around a circular cylinder: the static pressure is increased byapproximately the dynamic pressure of the wind at the stagnation point, and a muchstronger depression is prevailing at α ≈ 100º than that in the z = const. < zHE,top planes.This strong depression can not develop at the heat exchangers partly due to the towerdraught and also partly because of the braking effect of the ground.

Figure 4-13. Distribution of static pressure [Pa] at the tower outlet plane, wind speed: 6 m/s

Figure 4-13. shows the distribution of static pressure inside and outside the tower atthe tower outlet plane. The stagnation pressure in this plane is 16 Pa. This pressure forcesthe air to flow up and evade the front side of the upper edge of the tower, and allows the

Wind α

COMPUTATIONAL FLUID DYNAMICS MODELLING

72

exiting hot air jet to enter the ambience with a relatively low loss of energy. Inside theshell, in the first tierce of the cross-section, this results in a pressure less by some 5 Pathan the ambient pressure.

The depression evolving due to the velocity distribution around the cylindrical towershell is present intensively at this height, too, and causes static (over)pressure values of-18 ... -35 Pa in the rear two tierces of the outlet cross-section. The sideward depressionoutside the tower is so strong, that it sips out the hot air from the tower and induces twovortices rotating in the opposite directions to each other by forcing this hot air massdownwards outside the tower. These vortices are intensified by the strong thermalbuoyancy developed in the plume (which is deflecting to the horizontal direction),resulting in an upwelling velocity behind the tower even greater than the exit velocityfrom the tower, so between the two vortices the cold ambient air is carried up into theplume at the rear side of the shell.

Figure 4-14. shows the velocity vectors in a vertical plane which is perpendicular tothe wind direction and located 50 m behind the centreline of the tower for 6 m/s windspeed (wind is blowing towards positive x direction). Two vortices can be observed inFigure 4-14. flowing like two rolling gear-wheels.

Figure 4-14. Velocity vectors coloured by velocity magnitude [m/s] in the x = 50 m plane,wind speed: 6 m/s

In Figure 4-15. the same effects are illustrated in the x = 0 m plane, which isperpendicular to the wind direction.

In Figure 4-16. we can see the velocity distribution from the direction of wind. It iswell observable that according to the typical flow pattern around a cylinder, a velocitymagnitude of 13.8 m/s evolves beside the shell. It is important to note, that a highvelocity value is prevailing also in the region near by the plane of the outlet, while at thebottom, below the widening, the velocity is much less (11 m/s).

Wind

COMPUTATIONAL FLUID DYNAMICS MODELLING

73

Figure 4-15. Velocity vectors coloured by velocity magnitude [m/s] in the x = 0 m planeat wind velocity of 6 m/s

Figure 4-16. Contours of velocity magnitude [m/s] in the x = 0 m plane at wind velocity of 6 m/s

In Figure 4-17. it is illustrated that hot air is drawn out from the tower in theneighbourhood of the heat exchangers located at α = 90º. This phenomenon appearedalready at 9 m/s wind speed, but in a much less extent. The region influenced by thiseffect of the flow pattern around the tower did not increase at wind speeds higher than12 m/s. It is also important that at higher wind speeds hot air did not started to flow out ofthe tower at the leeward area (α = 180º).

COMPUTATIONAL FLUID DYNAMICS MODELLING

74

Figure 4-17. Contours of static temperature [K] in the z = 12 m plane at wind velocity of 12 m/s

Figure 4-18. Penetration of cold ambient air inside the tower (Iteration number: 790, cf. with Figure 4-9.)

It was already mentioned that at 1.5 m/s wind speed a very unstable tower operationwas experienced. The temperature field in the y = 0 m plane was displayed at everysecond iteration during the calculation, and a video file was created from these pictures.Based on this animation the instability of the air mass flow rate at the tower exit planecan be explained by the penetration into and subsequent upward ejection of cold ambientair out of the tower ever and anon. Figure 4-18. shows an individual static image from theanimation.

Windα

Wind

β

COMPUTATIONAL FLUID DYNAMICS MODELLING

75

Regarding Figure 4-18., one of my colleagues shared his experiences with us in EGIoffice (CSABA, 2003):

„I had two experiences from my youth regarding to this picture. One happened atVisonta on a foggy winter day in 1972, at the unit IV cooling tower of the present MátraPower Company. I was standing in the middle of the cooling tower with a seniorcolleague, and we suddenly observed that humid air was tumbled in at the top of thetower, then a cold waft flapped us after 15-20 seconds. We continued observing thisphenomenon, which occurred periodically over and over again.

The other happened during a measurement in Ibbenbüren in 1974, when I was able tolook inside a lower wet tower from the top of a dry cooling tower which was standingbeside the wet one. I observed the same as in this animation. The cold air tumbled intothe wet tower from time to time, hereby the fog disappeared in the tower and I could seethe upper packing from above. However, the air flow did not cease for any moment, thevapour appeared above the packing again and the tower was being filled up with itcontinuously. After a short time the phenomenon started from the beginning.”

We pointed out that this phenomenon may occur more possibly with low heat load andchoked louvers in the reality. In this case, the stability of the exit air mass flow is notsufficient, and the cold air tumbles into the tower. However, the operation of the naturaldraught tower will not stop because the ingressing cold air decreases not only the draught,but also the exit loss to such an extent that the exit loss becomes negative. The depressionraised by the wind begins to dominate, and this can restore the normal flow pattern veryquickly. Furthermore, starting the natural draught cooling tower is facilitated by wind,too.

4.5.2. Wind effect curve

In the first phase of the CFD calculations, eight different stationary wind speeds wereinvestigated between 0.3 and 20 m/s related to 10 m height above ground and 700 mupstream of the tower, while the tower ITD was held constant at its nominal value. As re-sult, the heat rejections of the tower for different wind speeds were obtained. In definingthe wind velocity profile at the inlet boundary of the domain, two wind profile exponentswere used: 0.087 (with a roughness height of 0.05 m of the ground) and 0.102 (with aroughness height of 0.4 m of the ground). This is illustrated in Figure 4-19., where v∞denotes the wind velocity at 10 m level above ground at the inlet boundary of the domain.

There is very little difference between the two curves. The largest deviation is 1.2 % at9 m/s. Till 6 m/s the curve relating to 0.05 m roughness height of the ground is above theother curve. However, the curve relating to 0.4 m roughness height resulted in highertower heat dissipations at wind speeds higher than 6 m/s.

Using the basic diagram of the cooling tower, ( )target,ϑwmfQ && =∗ related to design

ambient air dry bulb temperature, Tan, barometric pressure, pbn and nominal cooling waterflow rate, the results presented in Figure 4-19. could be transformed to nominal heat

COMPUTATIONAL FLUID DYNAMICS MODELLING

76

dissipation with increased tower ITD values. In this way, the „preliminary” wind effectcurves showing the relationship of the change in tower ITD value related to the nominaltower ITD, Δϑt/ϑt0, in function of the wind speed, v∞, were obtained. Since the„preliminary” wind effect curve with 0.05 m roughness height of the ground provided abetter approximation of the theoretical wind effect curve determined from the full-scalemeasurements, it was decided that further calculations should have been made with thisroughness height.

75

80

85

90

95

100

0 4 8 12 16 20w∞ [m/s]

Q [

% ]

Ground roughness height: 0.05 m

Ground roughness height: 0.4 m

Figure 4-19. Heat rejections of the cooling tower in case of fixed tower ITD value

In the second phase of the CFD calculations the tower ITD values were increased inFLUENT to such values that the resulting heat rejections of the tower were equal to thenominal heat dissipation for every investigated wind speed. In fact, the effect of wind onnatural draught cooling towers cannot be analysed correctly without taking into accountthe interaction of the tower with the other components of the power plant (e.g. the steamturbine). Although with increasing tower ITD (which means higher condensertemperature and a higher turbine back pressure) the cooling demand of the steam cycleslightly increases, it is reasonable to assume that the heat rejection is nearly constantwhen the tower ITD is increasing due to the wind. Because of the large number of theheat exchanger zones, the specification of the cooling water inlet temperature wasautomated by running journal files which contained all the necessary commands for thisprocedure and could be simply edited in Microsoft Excel. The resulting „final” windeffect curve is shown in Figure 4-20. Here the Δϑt/ϑt0 values are plotted in function ofthe wind speed va prevailing at the location of the anemometer during the full-scalemeasurements. In this manner the differences between the measured wind velocities andthe velocities in undisturbed flow (e.g. at 700 m upstream the tower at the inlet boundaryof the FLUENT model) were eliminated, if there were any. The wind effect curves fromthe measurement and FLUENT are in very good agreement with each other, themaximum deviation is 1.64 %.

v∞ [m/s]

COMPUTATIONAL FLUID DYNAMICS MODELLING

77

0.00

0.04

0.08

0.12

0.16

0.20

0 4 8 12 16 20wa [m/s]

Δϑ

t/ ϑt0

Theoretical wind effect curve, measurement

Theoretical wind effect curve, FLUENT

Figure 4-20. Wind effect curves for nominal operating conditions

Based on the above, it can be concluded that the CFD model is capable to predict thebehaviour of the natural draught dry cooling tower in wind.

4.5.3. Thermal performance

The thermal performance is reported for individual heat exchanger zones separately inFLUENT. Since there is a large number of cooling columns, the reporting wasautomated. First, journal files were read in, which contain the commands for theevaluation, whereupon the reported values were printed in the console window ofFLUENT. These data should have been copied into simple text files through theclipboard. After that these text files were opened in Excel, the evaluation was performedimmediately by using the previously adjusted referencing of the relevant cells containingthe heat rejection values of the individual heat exchanger zones.

The smallest wind velocity for which calculation was made, was 0.3 m/s at 10 m aboveground. In this case the heat rejected by the heat exchanger zones was nearly constant –the largest difference in the heat rejection among the cooling columns was 2.72 %.Therefore, the results for this wind speed were considered as the reference case, i.e. thewind-off conditions were approximated by this situation.

Figure 4-21. illustrates the heat rejection by the cooling columns spread out round thetower. The location of the largest heat dissipation can be found in the region of the 22nd

cooling column. Here also the air mass flow through the heat exchanger is the largest.The reason for this is that at this location the air flow must not change its directionsignificantly to pass through the heat exchanger (the wind direction is approx.perpendicular to the plane of the heat exchanger) and the velocity distribution is probablymore uniform along the width of the cooling column than at other locations.

The heat rejection is only slightly less at the rearward cooling deltas in comparisonwith the frontal ones. The smallest heat dissipation can be observed at coolers with

va [m/s]

COMPUTATIONAL FLUID DYNAMICS MODELLING

78

positions at the endpoints of the tower diameter perpendicular to the wind direction,because here the wind speeds up intensively, and therefore, the static pressure is low infront of the coolers. This small pressure difference between inside and outside the towerresults in low air mass flow through the heat exchangers. Also here we can find thelargest difference between the rejected heat of two adjacent cooling columns within acooling delta. This can be explained so that the dynamic pressure of the incoming air isutilised effectively by the heat exchanger facing frontally to the upstream flow, while itsneighbour does it barely due to the evolved vortices. The highest deviation as comparedto the nominal heat rejection of a cooling column is cc. 79.6 % between the 22nd (α = 30º)and the 73rd (α = 99.5º) heat exchanger zones. This disparity plays a significant role in thedecrease of the tower performance in wind.

0.0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Heat dissipation in case of 0.3 m/s wind speed

Figure 4-21. Dissipated heat (Q) in the cooling columns, wind speed: 6 m/s

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Heat dissipation in case of 0.3 m/s wind speed

Figure 4-22. Dissipated heat (Q) in the cooling columns, wind speed: 16 m/s

COMPUTATIONAL FLUID DYNAMICS MODELLING

79

Figure 4-22. shows the heat rejection of individual cooling columns for a higher,16 m/s wind speed. It can be seen that the heat exchanger zones between α = 72.3º ... 120ºhave very low heat dissipation values, which is caused by the high outer wind velocitiesat this part of the tower circumference, i.e. the pressure outside the tower is so low, thathot air is drawn out from the tower. This phenomenon decreases the natural draught andthe air mass flow rate of the tower.

0

60

120

180

240

300Average outlet watertemperatureValues lower thanaverageValues greater thanaverage

Wind direction

Cooling Tower Perimeter, α [ o ]

0oC

0

-2oC

+2oC

Figure 4-23. Outlet water temperatures averaged for every four neighbouring heat exchanger zones(cooling delta pairs), wind speed: 4 m/s

The outlet water temperature variation along the cooling tower perimeter plotted inFigure 4-23. confirms the calculated heat rejection values in the heat exchanger zones.Where there is higher heat dissipation, the cooling water outlet temperature iscorrespondingly lower. By FLUENT we obtained symmetrical distribution inFigure 4-23., due to the applied symmetry boundary condition. However, in real coolingtowers this figure may be slightly asymmetrical due to the surrounding buildings of thepower plant (cf. with Figure 2-3.).

4.5.4. Dimensionless numbers of flow similarity

Considering a typical case of 6 m/s wind speed at 10 m height above ground inundisturbed flow, and taking the arithmetic mean of cooling tower base and throatdiameters as the characteristic length,

( ) m4.9467121.821 =+⋅=L ,

the following dimensionless numbers can be calculated:

Reynolds number ≡ ratio of inertial forces to viscous forces,

35,201,989101.6094.946Re 5-

10 =⋅

⋅=

⋅=

νLv ,

Froude number ≡ ratio of inertial forces to body forces,

α [ º ]

Wind direction

COMPUTATIONAL FLUID DYNAMICS MODELLING

80

0.1974.9481.9

610 =⋅

=⋅

=Lg

vFr ,

Grashof number ≡ ratio of buoyancy forces to viscous forces,

( )( )

1525

3

2

3

10933.110609.1

4.9415-33.810.00322481.9⋅=

⋅

⋅⋅⋅=

⋅Δ⋅⋅=

−νβ LTgGr ,

Prandtl number ≡ ratio of momentum diffusivity to thermal diffusivity,

71.0Pr = ,

Rayleigh number, which is a measure of the strength of buoyancy-induced flow innatural (free) convection,

1515 101.37271.010933.1Pr ⋅=⋅⋅=⋅=GrRa ,

the Archimedes number, which is a combination of Grashof and Reynolds numbers,

( ) 6.161.148

4.941.113-1.18381.9Re 22

102 =

⋅⋅⋅

=⋅⋅Δ⋅

==v

LgGrArρρ ,

Peclet number for heat transfer,

24,993,41271.0989,201,35PrRe =⋅=⋅=Pe ,

and the Brinkman number, which indicates the importance of the thermal energycreated by viscous shear in the flow,

( )3-

2-5210 101.4

15-33.810.02546101.843

⋅=⋅

⋅⋅=

Δ⋅⋅

=T

vBrλμ .

The material properties of air were taken from RAŽNJEVIĆ (1964) at the temperatureaveraged between the ambient air and warmed air in the tower at design operatingconditions:

( ) ( ) C 24.40581.331521

2121 °=+⋅=+⋅= aa TTT

4.6. Grid adaption

In order to determine the sensitivity of the calculated results with the resolution of thegrid, an Y+ type grid adaption was made for the case with 0.3 m/s wind speed. With theoriginal grid the Y+ values ranged between 0 and 10,446.61. The logarithmic law of thewall is valid between 30 < Y+ < 500 (AZAD, 1993), so computational cells adjacent to allof the walls (excluding the exterior ground and the walls of the heat exchanger zones andtheir neighbouring walls) with Y+ values greater than 500 (altogether 29,641 cells) weremarked for hanging type refining. After that, the refinement process was restricted tocells with volumes greater than 0.025 m3. No limits were applied on the maximumnumber of cells generated during the adaption process, but the creation of new cells in theheat exchanger zones was precluded. (If cells at wall surfaces either neighbouring or

COMPUTATIONAL FLUID DYNAMICS MODELLING

81

within the heat exchanger zones were adapted, the calculation could not be continued,because an error occurred in FLUENT. Therefore, these cells were excluded from theadaption process.) With these restrictions the number of marked cells was decreased to29,092 (3.9 % of the total number of cells). The volume weight was set as 1.

The adaption process lasted 1 hour and 14 minutes, and the number of cells wasincreased by 27.1 %. This caused a 40 % increase in CPU time per iteration, and theresulting heat rejection of the cooling tower decreased by 0.08 % as compared to thatobtained by the original grid.

Considering the increased CPU time and the marginal change in tower heat rejectionafter adaption, it was concluded that the original grid was fine enough for our purposes.

4.7. Data about the computer hardware

The CFD calculations required much time. The necessary number of iterations forobtaining a converged solution ranged from 925 to 4,070 (on the average 2,421iterations). Therefore, a new computer configuration was purchased for thesecalculations, which was the state-of-the-art in the category of the Personal Computers inMay, 2002. With this hardware one iteration step lasted 55 s (which means 1 day and 13hours in CPU time for 2,421 iterations). The elements of this configuration are listedbelow.

Central Processing Unit Intel P4 2AGHz 478PGA

Motherboard ASUS P4T-E

Random Access Memory 4×512MB RAMBUS ECC

Video Card ASUS V8170 GeForce4 MX440 64MB

Hard Disk Drive Maxtor 740X 40GB IDE HDD 7200 rpm

Compact Disk Reader-Writer LG 16/10/40 IDE CDRW

Networking Card 3Com 905CTX-M 10/100 Mbit Lan Adapt

Tower Case DTK CAS-WT-PT074W

Monitor ELSAT-20H03T 20”

Operating System Windows 2000 Professional HU

Keyboard PS2 HU

Mouse SafeWay 4D Optical Scrolling Mouse USB & PS/2

Floppy Disk Drive 1.44MB FDD

WIND EFFECT IMPROVEMENT POSSIBILITIES

82

5. WIND EFFECT IMPROVEMENT POSSIBILITIES

The performance deterioration of the cooling tower in wind can be improved in severalways. The measures which were investigated in our survey can be grouped mainly intotwo groups: interventions on the water side and on the air side.

5.1. Measures on the water side

On the water side, two ideas were investigated for improving the wind effect. First, thetotal cooling water flow rate was increased with uniform distribution of flow rate amongthe heat exchangers. Second, the original amount of water flow rate to the tower was keptconstant but the distribution of the water flow rate among the heat exchanger sectors wascontrolled. In the above two cases the geometrical structure of our CFD model was notchanged.

5.1.1. Increasing the water flow rate to the tower

The modifications which were made to the reference CFD model introduced in Chapter 4are summarised below:

the original cooling water flow rate was quintupled in the heat exchanger zones,

in order to obtain the required heat rejection, the tower ITD value was decreasedalso for 0.3 m/s wind speed (ϑt0 = 20.82 ºC),

due to the increased water flow, the water side heat transfer coefficient was taken asthe treble of that in the Bursa model (partly, the aim in this step was to model theflow of steam in the cooler tubes instead of water). As a consequence, the heatexchanger effectiveness of both the main cooler and peak cooler heat exchangerswas changed.

Increasing the cooling water flow rate to such an extent is technically not realistic, butwith this theoretical simulation we can imagine how much the wind effect could beexpected to decrease by measures at the bottom part of the tower.

The main effect of the higher water flow rate is that in case of unchanged heatrejection the water temperature drop (cooling range) in the heat exchangers is less thanthat in the model of Bursa cooling tower. It is the reason also for the lower tower ITDvalue in no-wind case. The results proved that with higher water flow rate the distributionof cooling water outlet temperature around the tower perimeter was not distorted due towind so much as in the case with original water flow rate. According to Figure 5-1., itcan be assumed that where the outlet water temperature is lower than the average value(α ≈ -55º ... +55º and α ≈ 150º ... 210º) the heat transfer is enhanced in the case ofincreased water flow as compared to the case with original water flow, because the curvein cyan colour represents higher temperatures than the dashed curve in blue. With higher

WIND EFFECT IMPROVEMENT POSSIBILITIES

83

outlet water temperatures the temperature differences (or „driving potential”) between thecooling air and the water are higher in the heat exchangers. The situation is conversewhere the outlet water temperature is higher than its average value, i.e. the heat transfer ismore efficient in the case of original water flow rate.

0

60

120

180

240

300

Average outlet watertemperatureValues lower than average,original water flowValues greater than average,original water flowValues lower than average,increased water flowValues greater than average,increased water flow

Wind direction

Cooling Tower Perimeter, α [ o ]

0oC

0

-2oC

+2oC

Figure 5-1. Outlet water temperatures averaged for every four neighbouring heat exchanger zones (coolingdelta pairs), wind speed: 4 m/s

In Figure 5-2. the above statement is supported well. Where the outlet watertemperature is lower than its average value, the heat dissipation is greater in the case ofincreased water flow as compared to the case with original water flow. Where the outletwater temperature is over its average, there are smaller heat rejections in the case ofincreased water flow.

0.0

0.5

1.0

1.5

2.0

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Heat dissipationoriginal water flow

Heat dissipationincreased water flow

Heat dissipation in case of0.3 m/s wind speed

Figure 5-2. Dissipated heat (Q) in the cooling columns, wind speed: 4 m/s

We can conclude that at moderate wind velocities there are heat exchangers with bothenhanced and decreased heat transfer conditions. The wind effect curve of the coolingtower with increased water flow does not deviate significantly from that of Bursa tower.

WIND EFFECT IMPROVEMENT POSSIBILITIES

84

However, at higher wind speeds (> 8 m/s) the heat rejection of some coolers (α ≈ 72º ...128º) is decreased almost to zero, so their performance could not be even worse (seeFigure 5-3.). It means that the difference between the cases with increased and originalwater flow rate will comprise only the enhanced heat transfer at α ≈ 0º ... 50º and as aconsequence, the wind effect curve will be more advantageous for the increased waterflow. This can be observed in Figure 5-4., wherein the calculations were performed forthe same wind speeds as in the case of Bursa tower and the total heat dissipation was heldconstant. The higher water flow rate increases also the thermal inertia of the coolingsystem against wind-gusts.

0.00.51.01.52.02.53.03.54.0

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Heat dissipationoriginal water flow

Heat dissipationincreased water flow

Heat dissipation in case of0.3 m/s wind speed

Figure 5-3. Dissipated heat (Q) in the cooling columns, wind speed: 16 m/s

0.00

0.04

0.08

0.12

0.16

0.20

0 4 8 12 16 20wa [m/s]

Δϑ

t/ ϑt0

Increased cooling waterflow rate

Original cooling waterflow rate

Figure 5-4. Wind effect curves

5.1.2. Controlling the water flow into different heat exchanger sectors

Since the cooling deltas represent significant flow resistances in the cooling water circuit,the water flow rate can be considered nearly equally distributed among them. The mainidea of this measure was to increase the water flow rate in heat exchangers where the heat

va [m/s]

WIND EFFECT IMPROVEMENT POSSIBILITIES

85

rejection is increased due to wind. Consequently, the water flow should be decreased (e.g.with throttling by a control valve) in heat exchangers with lower heat dissipations in or-der to keep the total water flow rate of the tower unchanged. In this analysis the hydrauliccharacteristics of the cooling system was not taken into account, but it should be notedthat if the flow is distributed unevenly, the total water flow rate will slightly decrease.

It is technically not feasible to install a control valve for every individual heatexchanger. The main coolers are connected in parallel into six sectors on the water side,and usually one electromechanically actuated butterfly valve is built into both the sectorinlet and outlet pipelines of the realised cooling towers. Therefore, it is practical toinvestigate the effect of varying the flow distribution among the main cooler sectors. Theorientation of the sectors is illustrated in Figure 5-5. The water flow rate in the peakcooler sectors was not changed and the simulations were performed for only two windspeeds: 6 m/s and 16 m/s.

Figure 5-5. Orientation of main cooling sectors

The analysis was performed in four steps. The first unequal distribution of water flowrate was determined according to the following relation:

( )06

1,0,6

1

,0,,1 m

TT

TTm

iioutin

ioutini && ⋅

−⋅

−=

∑=

(5.1)

where im ,1& is the new water mass flow rate of the ith sector for the first calculationwith uneven flow distribution,

Tin is the inlet water temperature,

Tout,0,i is the outlet water temperature of the ith sector,

0m& is the original water mass flow rate of the sectors.

In equation (5.1) the values of Tin, Tout,0,i and 0m& were taken from the model of theBursa cooling tower with uniform flow distribution.

Wind direction

6

1

2

3

4

5

WIND EFFECT IMPROVEMENT POSSIBILITIES

86

The flow rate to the individual heat exchanger zones were adjusted by journal files inFLUENT, based on their location (sector number) around the tower perimeter. Afterobtaining a converged solution, it was found that the total heat dissipation of the towerwas increased (the tower ITD was kept unchanged). In the second step, by using theresults of this first calculation with uneven flow distribution, the flow rates weredetermined once more according to the following relation:

( ) ( )( )

( ) ( )( )

1

6

16

1,0,6

1

,1,,0,61

6

1,0,6

1

,1,,0,0,2

−

=

==

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛

∑ −⋅

−⋅−⋅⋅

−⋅

−⋅−⋅= ∑

∑ i

iioutin

ioutinioutin

iioutin

ioutinioutini

TT

TTTT

TT

TTTTmm && (5.2)

here im ,2& is the new water mass flow rate of the ith sector for the second

calculation with uneven flow distribution,

Tout,1,i is the outlet water temperature of the ith sector obtained from the firstcalculation with uneven flow distribution.

The flow rates determined in the second step deviated only slightly from thosedetermined by equation (5.1), which resulted in additional marginal increase in the totalheat dissipation of the tower (Q = 425.96 MW). By a third simulation it wasdemonstrated that when the inequality among the sector flow rates was carried too far, thethermal performance of the tower somewhat decreased.

Finally, in the fourth step, using the flow rates determined by equation (5.2) in thesecond step, it was found that the tower ITD could be decreased by 0.22 ºC in order toobtain the original heat dissipation of the tower for 6 m/s wind speed. This improvementin cooling tower heat dissipation efficiency could be attributed to the uneven water flowdistribution among sectors. For this case the heat rejections, water mass flow rates andoutlet water temperatures are illustrated in Figures 5-6. and 5-7., respectively.

0.0

0.5

1.0

1.5

2.0

2.5

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

1000

1500

2000

2500

3000

3500W

ater

flow

rat

e [k

g/s]

Heat dissipation in case of0.3 m/s wind speed

Water flow rate

Figure 5-6. Dissipated heat (Q) in the cooling columns and water mass flow rate of sectors,wind speed: 6 m/s

WIND EFFECT IMPROVEMENT POSSIBILITIES

87

Wind direction

0oC

0+3oC

-3oC

Figure 5-7. Deviation of outlet water temperatures averaged for every two neighbouring heat exchangerzones (cooling deltas) from their average value, wind speed: 6 m/s

The same calculation procedure for 16 m/s wind speed showed that due to theunevenly distributed cooling water mass flow rate, the total heat dissipation of the towerincreased to 464.02 MW and in the fourth step the tower ITD deterioration caused bywind could be decreased by 2.68 ºC.

5.2. Measures on the air side

Measures on the air side were investigated at the air inlet and outlet opening of the tower.At the air inlet area several ideas were analysed, while at the outlet only the effect of theratio of tower height to throat diameter was determined.

5.2.1. Measures at the air inlet opening

First without changing the geometry, a wind effect improvement possibility by means oflouver control was examined. After that the flow field was influenced by obstructivestructures in order to increase the air flow through the tower.

5.2.1.1. CONTROLLING THE LOUVERS

The mechanism of this measure can be explained so that at the windward side of thecooling tower (at sector 1 in Figure 5-5.), a significant amount of ambient air enters thetower and its temperature is not increased to such an extent as in another regions wherethe air flow rate is lower through the heat exchangers. This deteriorates the naturaldraught of the tower, because air masses of lower temperature are present inside thetower. Therefore, it can be suggested that with partially closing the louvers of sector 1,the tower heat dissipation efficiency may increase.

The calculations were made for 6 and 16 m/s wind speeds and the tower ITD valueswere taken from the wind effect curve in Figure 4-20. for the respective wind velocities.

WIND EFFECT IMPROVEMENT POSSIBILITIES

88

As a result, we could see the variation of the tower heat rejection in function of thepressure-jump coefficient, C2 (see Figure 5-8.).

410

415

420

425

430

435

440

445

0 100 200 300 400 500C2

Qto

wer

[MW

]

wind speed: 6 m/s

wind speed: 16 m/s

Figure 5-8. Tower heat rejection (Qtower) in function of the pressure-jump coefficient of sector 1

The pressure drop through the louvers is calculated according to equation (4.20). Byclosing the louvers, i.e. increasing the value of C2 in the sector facing directly the wind,first a slight increase was detected in tower thermal performance, then by closing furtherthe louvers, the thermal performance began to decrease. The maximum gain in towerperformance was not significant for 6 m/s wind speed (Q = 422.42 MW), but some full-scale site observations indicated higher improvement. In case of 16 m/s wind speed, thetower heat rejection increased to 439.04 MW.

From the adjusted pressure loss coefficient at the louvers in the FLUENT model andthe louver characteristics obtained by earlier experiments, I determined the requiredclosing angle of the louvers causing the maximum thermal performance improvement.This closing angle was in good accordance with the full-scale experience.

5.2.1.2. BAFFLE PLATES

The simulations to be introduced in this section required first of all the modification ofthe geometry. It means that additional surfaces were incorporated in the CFD model torepresent the deflector walls.

As the geometry was changed, some volumes should have been re-meshed, so thenumerical mesh was also altered. Therefore, in the first step some calculations were madeto reproduce the results obtained earlier with the original grid (Table 5-1.). The differencebetween the resulting tower heat rejection obtained with the two grids for 0.3 m/s windspeed is small and can be accepted:

%55.0%100421.15

418.84-421.15%100 =⋅=⋅−

old

newold

QQQ (5.3)

WIND EFFECT IMPROVEMENT POSSIBILITIES

89

Original model New gridWind speed [m/s]

Q [MW] ϑt [K] Q [MW] ϑt [K]Remark

421.15 24.52 418.84 24.52 Without baffle plates0.3

- - 418.67 24.52 With all baffle plates „on”6 421.19 25.51 418.66 25.56 Without baffle plates

Table 5-1. Preliminary calculation results with the new grid

With the new grid the influence of the baffle plates on the tower heat rejection wasexamined in the case of 0.3 m/s wind speed. It was found that when all baffle plates wereset as walls the thermal performance of the tower decreased barely, therefore the effect ofthe obstructive surfaces can be neglected in calm.

With the new grid and baffle plates set as internal surfaces the cooling tower required atower ITD value of 25.56 ºC in order to maintain the heat rejection of 418.66 MW in thecase of 6 m/s wind speed. The subsequent simulations were performed for 6 m/s windspeed with eight different configurations altogether, wherein the tower ITD value waskept unchanged (25.56 ºC) and the deviation of the resulting heat dissipations from418.66 MW showed the effectiveness of the actual wind effect improving structure.

The eight cases are described shortly below:

Case 1: 1.05 m long and 19.318 m high baffle plates are placed inside every coolingdelta module (comprising two adjacent cooling columns and a louver field). Theplanes of the baffle plates are aligned with the symmetry planes of the triangles ofthe respective cooling deltas.

Case 2: 2.1 m long baffle plates are placed in the symmetry plane of every coolingdelta module. The baffle plates sever entirely the volume inside the cooling deltasand their height is equal to the height of the heat exchangers, i.e. 19.318 m.

Case 3: 1.42 m long and 19.318 m high baffle plates are placed between theadjacent cooling deltas inside the cooling tower.

Case 4: the same arrangement as in Case 3 but with 2.13 m long baffle plates.

Case 5: 2.13 m long baffle plates are placed between the cooling deltas radiallyoutwards of the tower. The baffle plates extend from the ground to the top level ofthe heat exchangers.

Case 6: 2.1 m long baffle plates are placed radially outside the tower at the middlepoint of the louver field of every cooling delta. The baffle plates extend from theground to the top level of the heat exchangers.

Case 7: as a matter of fact the „baffle plates” are represented by 2 × 4 pieces ofcooling deltas outside the tower placed at α = 90º and α = 270º. These coolers areconsidered to be connected in parallel with the adjacent main cooling delta sectoron the water side when there is wind. The water flow rate is assumed to be uniform-ly distributed among the outer and the sector coolers, so that the cooling columns insectors 2 and 5 have lower cooling water flow rate than those in the other sectors.

WIND EFFECT IMPROVEMENT POSSIBILITIES

90

Case 8: 12 pieces of 16.27 m long baffle plates are arranged radially outside thetower. The height of the baffle plates is 21.68 m, i.e. they extend from the ground tothe top level of the heat exchangers.

Figure 5-9. Schematic drawings of the investigated configurations

Wind direction

Case 8Case 7

Wind directionα

21

546

3

Case 6Case 5

Case 4

Case 2

Case 3

These are cooling deltas

Case 1

Louvers Cooling columns

WIND EFFECT IMPROVEMENT POSSIBILITIES

91

Cases 1-6 are expected to decrease the heat rejection differences between the adjacentcooling columns observed in wind. Cases 7 and 8 involve larger but fewer obstructivesurfaces and are aiming at influencing the wind-induced flow pattern around the tower.

The tower heat rejection values obtained from the simulations (with ϑt = 25.56 ºC and6 m/s wind speed) are summarised in Table 5-2. The case number „0” indicates theconfiguration without any baffle plate.

Case No. Q [MW] ΔQ [MW] dQmax [MW]

0. 418.66 0.00 0.7021. 419.70 1.04 0.8822. 428.05 9.39 0.2883. 423.15 4.49 0.5934. 425.85 7.20 0.5895. 436.33 17.67 0.1446. 436.81 18.16 0.1547. 440.25 21.59 1.6058. 447.11 28.46 0.785

Table 5-2. Results of the simulations

We can see that every configuration yielded an improvement. In Table 5-2. ΔQ is thedifference in tower heat rejection between the case in question and case number „0”,while dQmax is the maximum difference between the heat dissipation of neighbouringcooling columns for the case in point.

In Figure 5-10. the path lines and pressure contours are illustrated for Case 0 (withoutany baffle plates). We can see that if baffle plates corresponding to Case 1 were installed(shown with dashed line in Figure 5-10.), path lines number 2 and 3 would not flowthrough cooling column A but through column B. This results in an increase in dQmax forCase 1 as compared to that of Case 0. However, the tower heat rejection is higher inCase 1 than in Case 0. The reason for this may be that at locations around the towerperimeter other than α ≈ 90º more advantageous flow distributions may prevail for Case 1on the whole. In Case 2 the baffle plates are longer and enhance the air flow throughcooling column A, therefore dQmax decreases in this case. The baffle plates of Case 3make an end of path line number 1 which again results in a lower dQmax compared to thatof Case 0. Case 4 eventuates a slightly better dQmax value than Case 3. It was peculiar forCases 0-4 that cooling columns at positions similar to B in Figure 5-10. (heat exchangerzones with even ordinal numbers) had higher heat rejection than cooling columns atpositions similar to A (heat exchanger zones with odd ordinal numbers).

WIND EFFECT IMPROVEMENT POSSIBILITIES

92

Figure 5-10. Contours of static pressure [Pa] overlaid by path lines in the z = 12 m horizontal plane,in the region α ≈ 90º

From Table 5-2. we can see that Case 5 has the lowest dQmax value. Figure 5-11.shows that the distribution of the heat rejection among the cooling columns improvedsignificantly (cf. with Figure 4-21.).

0.0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Average heat dissipation

Figure 5-11. Dissipated heat (Q) in the cooling columns, wind speed: 6 m/s (Case 5)

In contrast to Cases 0-4, it was found that with the baffle plate arrangement of bothCase 5 and 6 the cooling columns with odd ordinal numbers have higher heat dissipationthan that with even ordinal numbers. In Figures 5-12. and 5-13. the path lines areillustrated through the heat exchanger zones for Case 5 and 6, respectively. In the regionα ≈ 90º, where dQmax has a high value, it can be observed that in Case 5 the air flow isdistributed more uniformly between the two adjacent cooling columns than in Case 6, and

Wind direction

1 2 3

AB

WIND EFFECT IMPROVEMENT POSSIBILITIES

93

this explains why the value of dQmax is higher for Case 6. Nevertheless, Case 6 yielded aslightly higher increase in the tower heat rejection than Case 5. It may be due to that atlocations around the tower perimeter other than α ≈ 90º more advantageous flowdistributions may prevail for Case 6 and this baffle plate arrangement provides morebenefits on the whole.

Figure 5-12. Path lines coloured by velocity magnitude [m/s] for Case 5 in the z = 12 m horizontal plane,in the region α ≈ 90º

Figure 5-13. Path lines coloured by velocity magnitude [m/s] for Case 6 in the z = 12 m horizontal plane,in the region α ≈ 90º

Wind direction

Wind direction

WIND EFFECT IMPROVEMENT POSSIBILITIES

94

0.0

0.5

1.0

1.5

2.0

2.5

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

1000

1500

2000

2500

3000

3500

Wat

er fl

ow r

ate

[kg/

s]Average heat dissipation

Water flow rate

Figure 5-14. Dissipated heat (Q) in the main cooling columns and water mass flow rate of sectors, Case 7

Figure 5-15. Contours of static pressure [Pa] in the z = 12 m horizontal plane, Case 7

With Case 7, the wall placed between the outer coolers and the tower peripheryresulted in a sharp decrease in the heat rejection of the main cooling columns afterα = 90º (Figure 5-14. and 5-15.). Practically, these heat exchangers can receive only thewarm air exiting from the outer coolers. Therefore, I have removed this connecting plateleaving a 2.06 m wide gap instead and run the simulation again. As a result, the towerheat rejection and dQmax were obtained as 440.36 MW and 0.959 MW, respectively. Thearrangement investigated in Case 7 can be applied effectively at sites where the windalways blows from the same direction.

The highest tower heat rejection was obtained in Case 8. Figures 5-16. and 5-17. showthat the influence of the baffle plates on the flow field is significant. It can be also seenthat in the leeward region of the wind-walls the heat exchangers operate with decreased

Wind direction

Connecting plate

2.06 m

WIND EFFECT IMPROVEMENT POSSIBILITIES

95

air mass flow rate. I tried to improve this negative effect by replacing the wall boundarycondition at the end of the wind-walls adjoining the tower with porous faces havingdimensions 5.35 m in width and 21.68 m in height. However, by altering the pressure-jump coefficient so that the air velocity through the porous part of the wind-wall at α =90º was decreased from 7.1 m/s (this corresponds to a pressure-jump coefficient of zero)to 3 m/s, the change in tower heat rejection was between -1.16 ... 0.81 MW which meansthat in this manner the additional tower performance improvement is very low.

Figure 5-16. Contours of static pressure [Pa] in the z = 12 m horizontal plane, Case 8

0.0

0.3

0.6

0.9

1.2

1.5

1.8

2.1

0 30 60 90 120 150 180 210 240 270 300 330 360

α [ o ]

Q [M

W]

Average heat dissipation

Figure 5-17. Dissipated heat (Q) in the cooling columns, Case 8

5.2.2. Measures at the tower outlet

The final set of simulations was made with a modified shape of the tower shell. It meansthat the height and the throat diameter were changed from 135 m and 67 m to 127.1 m

Wind direction

WIND EFFECT IMPROVEMENT POSSIBILITIES

96

and 73.7 m, respectively. This is illustrated in Figure 5-18. The bottom part of thecooling tower (base diameter and position of the cooling deltas) was not altered, and themodified tower shell dimensions were determined by EGI’s internal sizing program so,that they must ensure the same tower performance (unchanged nominal tower ITD, heatrejection and air mass flow rate) in no-wind case.

With this larger tower throat diameter the air exit velocity and exit loss from the towerwill be both smaller, so the tower height could be decreased. However, due to the lowerair exit velocity the tower may be more sensitive to cross-winds. This problem will beanalysed in this section.

The simulations were performed using the nominal tower ITD value, so the resultingoverall heat rejections indicate the cooling capability of the tower. For 0.3 m/s windspeed, the tower heat rejection was higher by only 0.04 % than that obtained with theoriginal tower geometry. In case of 6 m/s wind speed the average tower air mass flow ratewas by as much as 4.77 % lower with the modified tower shell, however, the solution wasmuch less stable than with the original geometry – there were significant changes in theair mass flow rate through the tower even after 4,000 iterations (up to ±1,104 kg/sdeviations from the average value). It can be assumed that with lower air exit velocity thecold ambient air is more susceptible to penetrate into the tower in the outlet plane, andthis caused the aforementioned instability.

For higher wind speeds, the stability of the solution was acceptable, and the differencesbetween the heat rejections obtained with the original and modified geometry, using thenominal tower ITD value, were 3.258, -0.742, -2.187 and -5.646 MW in case of 9, 12, 16and 20 m/s wind speeds, respectively. (For these wind speeds the reference tower heatrejections with the original geometry were indicated in Figure 4-19. with 0.05 mroughness height of the ground).

Figure 5-18. Original and modified cooling tower shells

Original tower shell

Modified tower shell

WIND EFFECT IMPROVEMENT POSSIBILITIES

97

5.3. Economical evaluation

I have performed a simple economical evaluation of wind effect improving measures.The basis of this analysis was the loss of electricity generation caused by the effect ofambient wind on the cooling tower.

Hourly average meteorological records were available from the power plant’smeteorological station for one month period (June, 2001). By substituting these dataexcluding wind velocities into cooling system sizing calculations, I determined firstly thecondenser temperatures in case of no wind for the respective one-hour long periods. Thecondenser temperatures and steam turbine characteristics allowed the calculation ofelectricity generated by the steam turbine with the assumption of continuous 100 % heatinput into the steam cycle.

In the next step, with the help of hourly average ambient wind speed records and windeffect curves obtained from on-site measurement and CFD simulations, changes in warmwater temperature due to wind in the cooling tower and corresponding electricityproductions were calculated for the different wind effect improving measures. Theelectricity generation values were compared with those from no-wind case, and thedifferences during the one-hour long periods were summed to determine the monthlyelectricity losses for the investigated improving measures.

The annual costs occurring due to ambient wind were calculated with an electricityprice of 4 ¢/kWhe and a power plant load factor of 85 %. As in cold winter weathersituations the cooling air flow is throttled by the louvers of the tower anyhow in order toavoid condenser pressures below the choking point of the steam turbine, the wind effectcan be compensated by less throttling at the louvers. Therefore, annual costs werereduced by a factor of 0.9.

The results for the Bursa 1,400 MWe CCPP are shown in Table 5-3. Thesupplementary investment costs for baffle plates, Cases 1-8 were determined according togeneral pricing practice used at EGI - Contracting Engineering Co. Ltd. and their paybackperiods were calculated with an interest rate of 4 %.

Cost [$/a] Benefit [$/a] Supplementaryinvestment [$] Payback period [a]

Base case 371,420 0 - -Increased cooling water flow 299,902 71,519 - -Controlled cooling water flow 339,609 31,811 - -Controlled louvers 365,875 5,545 - -Baffle plates, Case 1 368,998 2,422 352,127 >100Baffle plates, Case 2 321,554 49,866 704,253 20Baffle plates, Case 3 366,334 5,086 476,209 >100Baffle plates, Case 4 341,531 29,889 714,314 65Baffle plates, Case 5 248,184 123,237 1,603,306 18Baffle plates, Case 6 244,016 127,404 1,580,724 17Baffle plates, Case 7 214,416 157,005 645,954 5Baffle plates, Case 8 156,755 214,666 974,454 5Modified tower shell 525,189 -153,769 - -

Table 5-3. Results of economical evaluation

SUMMARY OF NEW RESULTS

98

6. SUMMARY OF NEW RESULTS

Wind effect on natural draught cooling towers has complex physics. It takes place in theatmospheric boundary layer, and for a long time, the only reliable method to determine itsquantitative value has been the on-site measurement on full-scale towers.

Recently, we have seen a very rapid development in the field of Computational FluidDynamics (CFD), whereby simulation of fluid flows in a complex geometry becamefeasible. CFD modelling can serve as a powerful tool to predict the degree of wind effecton cooling towers already in the design stage, which can influence the selection of theoptimum tower design. When an air-cooled cooling tower is subjected to cross-wind, theflow field is distorted around the tower. As a consequence, some heat exchangers havehigher cooling air flow through them, while other coolers have lower. On the whole, theefficiency of the tower decreases, which means that the condensation heat from the steamcycle can be rejected to the environment only with increased cooling water temperaturesin the cooling tower. This results in a higher temperature of condensation in thecondenser and hereby a higher steam turbine back pressure, which cuts down the thermalefficiency and capacity of the power plant – the same output of electric energy willrequire more input of fuel, as well as the unit’s capacity, available for the electricity grid,decreases. The costs caused by the decreased cooling plant capability in wind can bedetermined, and these costs could make reasonable the application of such accessoriesthat can mitigate the wind effect economically.

Actual wind effects on specific cooling towers depend on a large number ofparameters, so the effect of wind is difficult to express in a general way. Therefore,within the frameworks of my Ph.D. work a specific cooling tower was selected to beanalysed. The methods included full-scale measurements in the field and state-of-the-artCFD investigations of the problems relating to the operation of the aforementionednatural draught power plant cooling tower in ambient winds. The applied ComputationalFluid Dynamics software was FLUENT.

In this Ph.D. dissertation the details of the computational model were described, theobtained flow fields were analysed, and the wind effect curve resulting from the CFDcalculations was compared with that obtained from full-scale site measurements. Theagreement between experimental and numerical results was more than satisfactory.

Some improving measures aimed at lowering the deteriorating effect of wind on thecooling tower thermal performance were also discussed. The numerical simulationsshowed moderate improvement possibilities, however, full-scale site measurements werenot performed yet for these cases.

The purpose of this Ph.D. study was to find and analyse new solutions for improvingthe behaviour of natural draught power plant dry cooling towers in wind. The mainresults are summarised below.

SUMMARY OF NEW RESULTS

99

6.1. Thesis 1.

Based on full-scale measurements on a given cooling system, I have determined therelationship between the ambient wind speed and the change in warm water temperatureentering the cooling tower. I have demonstrated by statistical analysis of the measuredwind velocities the sense and extent of the effect of wind-gusts occurring during the one-hour averaging period used in measurement evaluation.

6.2. Thesis 2.

I have created a mathematical cooling tower model more detailed than those known in theliterature, and I have determined the wind effect curve also by numerical simulations. Thefull-scale measurements confirmed the correctness of my model very well. Furthermore,the visualised scalar distributions, vector plots, flow animations and resulting numericalreports agree with every known or formerly supposed theory and experience. Beyond mysurveys presented in my Ph.D. thesis, my numerical model provides the basis to performfurther experimental studies in the future.

6.3. Thesis 3.

My investigations have shown that wind effect improving measures can be appliedalready at the cooling system’s design stage, too. I have carried out technical andeconomical analysis of two design alternatives different from usual practice: increasedcooling water mass flow rate and modified shape of tower shell (stubbier tower). With theenlarged cooling water flow the loss of electricity occurred due to the effect of windcould be decreased, which means 71,519 USD/a extra income for the power plantannually. The modified shape of the tower would yield extra profit only in case of higherwind velocities, so the extra income is -153,769 USD/a.

6.4. Thesis 4.

I have found that the effect of wind on already existing and operating cooling systems canbe improved by appropriate adjustment of the operational parameters. I have carried outtechnical and economical analysis of two interventions of this kind, namely non-uniformcooling water flow distribution among the heat exchangers according to their coolingduty perturbed by the wind as well as cooling air flow control by louvers. In both casesthe loss of electricity occurred due to the effect of wind could be decreased, which means31,811 USD/a and 5,545 USD/a extra income for the power plant, respectively.

SUMMARY OF NEW RESULTS

100

6.5. Thesis 5.

I have found that the effect of wind on already existing and operating cooling systems canbe improved also by the installation of windbreak walls requiring additional investments,and I have carried out technical and economical analysis of eight different wind-wallarrangements. With all configurations developed by me, the loss of electricity occurreddue to the effect of wind could be decreased. Hereby, extra income of between2,422 USD/a and 214,666 USD/a can be achieved in the power plant. Depending on thearrangement, the investment costs of the wind-walls ranged between 352,127 USD and1,603,306 USD. The two most advantageous arrangements have payback periods of 5years.

Budapest, 23 February 2005

……………………………….Nimród Kapás

Candidate for the Ph.D. degree

REFERENCES

101

REFERENCES

[1] AL-WAKED, R., BEHNIA, M. AND MADADNIA, J. (2001): Numerical Modelling and Validation ofNatural Draught Cooling Towers under Crosswind. Proceedings of the 12th IAHR Symposium inCooling Tower and Heat Exchangers, Editors: Jafar Madadnia & Homa Koosha, University ofTechnology, Sydney, Faculty of Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.:W-103, pp. 235-245. ISBN: 1-86365-697-9.

[2] ASME PERFORMANCE TEST CODE (PTC) 23-12986 (R1992): Atmospheric Water CoolingEquipment. American Society of Mechanical Engineering (ASME), United Engineering Centre, 345East 47th Street, New York, NY, 10017.

[3] AZAD, R. S. (1993): The Atmospheric Boundary Layer for Engineers. Fluid Mechanics and itsApplications, Volume 17. Kluwer Academic Publishers, Dordrecht, Boston, London, 1993. ISBN: 0-7923-2187-1.

[4] BAER, E. (1979): Einfluß des Windes auf die Strömung im Kamin eines Naturzugkühlturms.Fortschritt-Bericht der VDI-Zeitschriften, Reihe 6, Nr. 57. VDI-Verlag GmbH, Düsseldorf, 1979.(In German)

[5] BAER, E., ERNST, G. AND WURZ, D. (1980): Effect of Wind on the Flow within the Shell of aNatural Draught Cooling Tower. VGB KRAFTWERKSTECHNIK 60, Number 11, November 1980.

[6] BALOGH, A. AND TAKÁCS, Z. (2001): Field Measurements Verify Sizing Calculations in DryCooling Heller System. Proceedings of the 12th IAHR Symposium in Cooling Tower and HeatExchangers, Editors: Jafar Madadnia & Homa Koosha, University of Technology, Sydney, Facultyof Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.: T-103, pp. 116-124. ISBN: 1-86365-697-9.

[7] BANQUET, F. AND DELABRIERE, H. (1998): Code N3S-AERO pour la Modelisation desAéroréfrigérants Intérêt et Applications Industrielles. Proceedings of the 11th IAHR Cooling TowerSymposium. Cottbus, Germany. 22 September, 1998. (In French)

[8] BENDER, T. J., BERGSTROM, D. J. AND REZKALLAH, K. S. (1996): A Study on the Effects of Windon the Air Intake Flow Rate of a Cooling Tower, Part 2, Wind Wall Study. Journal of WindEngineering and Industrial Aerodynamics, Vol. 64, pp. 61-72.

[9] BERGSTROM, D. J., DERKSEN, D. AND REZKALLAH, K. S. (1993): Numerical Study of Wind Flowover a Cooling Tower. Journal of Wind Engineering and Industrial Aerodynamics 46 & 47. pp. 657-664.

[10] BERLINER, P. (1980): Gegenwärtige Kühlturmtechnik. Wärme Band, Heft 3, 1980. (In German)

[11] BLANQUET, J. C., GOLDWIRT, F. AND MANAS, B. (1986): Wind Effects on the AerodynamicOptimization of a Natural Draft Cooling Tower. 5th IAHR Cooling Tower Workshop, Monterey,California, 29 September – 3 October, 1986. International Association for Hydraulic Research,Division III – Industrial and Structural Fluid Mechanics, Section on Fluid Mechanics in EnergyProduction. Proceedings prepared for the Electric Power Research Institute.

[12] BOŠNJAKOVIĆ, F., VILIČIĆ, M. AND SLIPČEVIĆ, B. (1951): Einheitliche Berechnung vonRekuperatoren. VDI-Forschungsheft 432. Düsseldorf: VDI-Verl. 1951.

REFERENCES

102

[13] BOURILLOT, C. AND GRANGE, J. L. (1980): Influence du Vent, sur les Performances d’un RéfrigérantHumide à Tirage Naturel. SHF – Seizièmes Journées de l’Hydraulique, Paris, September 1980. (InFrench)

[14] BOURILLOT, C. (1983): TEFERI: Numerical Model for Calculating the Performance of anEvaporative Cooling Tower. EPRI, Special Report CS-3212SR, 1983.

[15] BUCHLIN, J. M. AND OLIVARI, D. (1982): Experimental Investigation of Flow in a Natural DraftCooling Tower. Von Karman Institute for Fluid Dynamics, Preprint 1982-33, August 1982.

[16] BUXMANN, J. AND HAMANN, H. (1974): Berechnung der Kennfelder von Trockenkühlern. BWK 26,H. 10., pp. 421-428, 1974.

[17] BUXMANN, J. (1977): Der Einfluß von Temperaturschichtungen auf die Wärmeleistung einesTrockenkühlturmes mit Naturzug. BWK 29, H. 3, S. 90/4, 1977.

[18] BUXMANN, J. (1986): Dry Cooling Tower Characteristics Effected by Cooling Circuit and the TopShape of the Tower. 5th IAHR Cooling Tower Workshop, Monterey, California, 29 September – 3October, 1986. International Association for Hydraulic Research, Division III – Industrial andStructural Fluid Mechanics, Section on Fluid Mechanics in Energy Production. Proceedingsprepared for the Electric Power Research Institute.

[19] CAYTAN, Y. AND FABRE, L. (1989): Wind Effects on the Performance of Natural Draught WetCooling Towers Comparison of Constructors’ Proposals and Realisation of Performance ControlTests. In: Dinelli, G. and Bellagamba, B. (Eds.): Proceedings of the International Cooling-TowerConference held in Pisa, Italy, 4-7 October, 1988. Prepared for ENEL, CRTN and EPRI; EPRI GS-6317, April 1989. pp. 4.1-1 – 4.1-17.

[20] CHABARD, J. P., NITROSSO, B. AND POT, G. (1993): Projet N3S de Mécanique des Fluides: ManuelThéorique des Version 3 et 4. Note EDF/DER HE-41/93/038B, Décembre 1993. (In French)

[21] CHILTON, H. (1952): Performance of Natural-Draught Water-Cooling Towers. Proc. I. Elec. Eng.,99. Part II, pp. 440-452.

[22] CHOLNOKY, P. (1989): Acélszerkezetű Hűtőtornyok Fejlődése és Gazdaságossága. Conference onCooling Systems with Cooling Towers. Budapest, 27-29 September, 1989. Published by theScientific Society for Building, 89/1136 MTESZ Házinyomda, Budapest. pp. 141-152. (InHungarian)

[23] CHRISTOPHER, P. J. AND FORSTER, V. T. (1969-1970): Rugeley Dry Cooling Tower System. Proc.Inst. Mech. Engrs 184 (11), 197-211.

[24] CSABA, G. AND KAPÁS, N. (2002): Heller Rendszerű Hűtőberendezések Méréskiértékelő Programja,a Bursa-i és az Al-Zara-i Garanciamérések Kiértékelése. Internal report 6810-ECSR-T10. EGI -Contracting Engineering Co. Ltd. 7th August, 2002.

[25] CSABA, G. (2003): Private Communication. EGI - Contracting Engineering Co. Ltd., Budapest,February 2003.

[26] DASHKOV, G. V., FISENKO, S. P., SOLODUKHIN, A. D. AND VLASOV, A. V. (1996): OpticalDiagnostics of Flow Swirling in Cooling Towers. ImechE Conference Transaction v3, pp. 565-573.

[27] DEMUREN, A. O. AND RODI, W. (1987): Three-Dimensional Numerical Calculations of Flow andPlume Spreading past Cooling Towers. J. Heat Transfer, 109. pp. 113-119.

[28] DERKSEN, D. D., BENDER, T. J., BERGSTROM, D. J. AND REZKALLAH, K. S. (1996): A Study on theEffects of Wind on the Air Intake Flow Rate of a Cooling Tower, Part 1: Wind Tunnel Study.Journal of Wind Engineering and Industrial Aerodynamics, Vol. 64, pp. 47-59.

REFERENCES

103

[29] DOSKEMPIROV, B. M., NIETKALIEV, S. ZH., ATAGELGIEVA, L. ZH., BAEKOV, E. S. AND HASENOV,B. O. (1992): Aerodynamical Researches of Cooling Tower Operating. 8th IAHR Cooling Tower andSpraying Pond Symposium, Johann-Gottfried-Tulla-Hörsaal, Kollegiengebäude am Ehrenhof,Universität Karlsruhe, Karlsruhe, Germany, 5-9 October, 1992.

[30] DREYER, A. A., OOSTHUIZEN, P. C. AND VAN STADEN, M. P. (1998): The Effect of Wind on thePerformance of Counterflow Natural Draught Evaporative Cooling Towers. Proceedings of the 11th

IAHR Cooling Tower Symposium. Cottbus, Germany. 22 September, 1998.

[31] DU PREEZ, A. F. (1992): The Influence of Cross-Winds on the Performance of Natural Draft Dry-Cooling Towers. Ph.D. thesis, Dept. of Mech. Eng., University of Stellenbosch, Stellenbosch, 7600,Republic of South Africa, 1992.

[32] DU PREEZ, A. F. AND KRÖGER, D. G. (1992): Investigation into the Influence of Cross-Winds on thePerformance of Dry-Cooling Towers. 8th IAHR Cooling Tower and Spraying Pond Symposium,Johann-Gottfried-Tulla-Hörsaal, Kollegiengebäude am Ehrenhof, Universität Karlsruhe, Karlsruhe,Germany, 5-9 October, 1992.

[33] DU PREEZ, A. F. AND KRÖGER, D. G. (1993): Effect of Wind on Performance of a Dry-CoolingTower. Heat Recovery Systems & CHP, Vol. 13, No. 2, pp. 139-146. Pergamon Press Ltd.

[34] DU PREEZ, A. F. AND KRÖGER, D. G. (1994): The Influence of a Buoyant Plume on the Performanceof a Natural Draft Cooling Tower. International Association for Hydraulic Research – 9th CoolingTower and Spraying Pond Symposium. Von Karman Institute for Fluid Dynamics, Rhode-Saint-Genèse, Belgium. 20-23 September, 1994.

[35] DU PREEZ, A. F. AND KRÖGER, D. G. (1995a): The Effect of the Heat Exchanger Arrangement andWind-Break Walls on the Performance of Natural Draft Dry-Cooling Towers Subjected to Cross-Winds. Journal of Wind Engineering and Industrial Aerodynamics, Vol. 58, pp. 293-303, December1995.

[36] DU PREEZ, A. F. AND KRÖGER, D. G. (1995b): Effect of the Shape of the Tower Supports and Wallson the Performance of a Dry-Cooling Tower Subjected to Cross Winds. Heat Transfer Engineering,Vol. 16, pp. 42-49, February 1995.

[37] ERNST, G. AND WINKLER, C. (1994): Flow Patterns of Cooling Tower Plumes in the NearAtmosphere – Results from a Mathematical Model. International Association for HydraulicResearch – 9th Cooling Tower and Spraying Pond Symposium. Von Karman Institute for FluidDynamics, Rhode-Saint-Genèse, Belgium. September 20-23, 1994.

[38] ETZOLD, F. AND FIEDLER, S. (1976): The Near-Wake Structure of a Cantilevered Cylinder in a CrossFlow. Zeitschrift für Flugwiss Weltraum-forsch, Vol. 24, pp. 77-82.

[39] FABRE, L. AND LEGRAND, G. (1988): Mistral, New Test Bench for Cooling Tower Components(Nuclear Generation Centre of Bugey, France). 6th IAHR Cooling Tower Workshop, Pisa, Italy,October 4-7, 1988.

[40] FARIVAR, D. (1981): Turbulent Uniform Flow around Cylinders of Finite Length. AIAA Journal,Vol. 19, pp. 275-281.

[41] FLASSAK, T. (1990): Ein Nicht-Hydrostatisches Mesoskaliges Modell zur Beschreibung derDynamik der Planetaren Grenzschicht. VDI-Verlag, Düsseldorf, pp. 203.

[42] FLASSAK, T. AND MOUSSIOPOULOS, N. (1987): An Application of an Efficient Non-HydrostaticMesoscale Model. Bound. Lay. Met. 41, pp. 135-147.

[43] FLUENT (2001a): FLUENT 6 Getting Started. Lebanon (NH), Fluent Inc., 2001.

[44] FLUENT (2001b): FLUENT 6 Text Command List. Lebanon (NH), Fluent Inc., 2001.

REFERENCES

104

[45] FLUENT (2001c): FLUENT 6 Tutorial Guide Volume 1-2. Lebanon (NH), Fluent Inc., 2001.

[46] FLUENT (2001d): FLUENT 6 UDF Manual. Lebanon (NH), Fluent Inc., 2001.

[47] FLUENT (2001e): FLUENT 6 User’s Guide Volume 1-5. Lebanon (NH), Fluent Inc., 2001.

[48] FOURNIER, Y. AND BOYER, V. (2001): Improvements to the N3S-AERO Heat Exchanger andCooling Tower Simulation Code. Proceedings of the 12th IAHR Symposium in Cooling Tower andHeat Exchangers, Editors: Jafar Madadnia & Homa Koosha, University of Technology, Sydney,Faculty of Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.: M-103, pp. 31-38.ISBN: 1-86365-697-9.

[49] GELDENHUYS, J. D. AND KROGER, D. G. (1986): Aerodynamic Inlet Losses in Natural DraughtCooling Towers. 5th IAHR Cooling Tower Workshop, Monterey, California, September 29-October3, 1986. International Association for Hydraulic Research, Division III – Industrial and StructuralFluid Mechanics, Section on Fluid Mechanics in Energy Production. Proceedings prepared for theElectric Power Research Institute.

[50] GELGAND, R. E., SUKHOV, Y. A. AND SHISHOV, V. I. (1990): Laboratory and MathematicalModelling of Thermal Processes in Cooling Towers. International Assoc. for Hydraulic Research,7th Cooling Tower and Spray Pond Symposium. Leningrad, USSR, May 29-June 2, 1990.

[51] GOECKE, E. (1969): Die Kondensationsanlage des 150-MW-Blockes im Kraftwerk Ibbenbüren derPreußag AG. VIK-Ber. Nr. 176. Essen, 1969.

[52] GRANGE, J. L. (1994): Calculating the evaporated water flow in a wet cooling tower. 9th IAHRCooling Tower and Spraying Pond Symposium, Brussels, September 20-23, 1994.

[53] GRUBER, J. AND BLAHÓ, M. (1981): Folyadékok Mechanikája. Tankönyvkiadó, Budapest, 1981.

[54] HILL, W. (2001): Relating Power Station Efficiency to Cooling Tower Performance. Proceedings ofthe 12th IAHR Symposium in Cooling Tower and Heat Exchangers, Editors: Jafar Madadnia &Homa Koosha, University of Technology, Sydney, Faculty of Engineering, Sydney, Australia, 11-14November, 2001, Paper No.: M-107, pp. 60-63. ISBN: 1-86365-697-9.

[55] KAPÁS, N. (2002): Turbulens áramlások modellezése az atmoszférikus határrétegben, (Turbulentflow modelling in the atmospheric boundary layer), Magyar Energetika, Vol. 10, No. 6, pp. 31-35,December 2002. ISSN 1216-8599.

[56] KESZTHELYI, Á. (1996): Főhűtő Rendszer Hőtechnikai Számítás. Bursa 1400 MWe CCPP, IndirectDry Cooling Heller System. Internal Calculation, EGI - Contracting / Engineering Co. Ltd. EGIDoc. No.: 6524-LK-TM.01.00.

[57] LAUNDER, B. E. AND SPALDING, D. B. (1974): The Numerical Computation of Turbulent Flows.Comput. Methods Appl. Mech. Engng. 3, pp. 269-289.

[58] LAURAINE, H., LEMMENS, P. AND MONJOIE, M. (1989): Experimental Data coupling AtmosphericTemperature Inversions and Cooling Tower Performances. In: Dinelli, G. and Bellagamba, B.(Eds.): Proceedings of the International Cooling-Tower Conference held in Pisa, Italy, October 4-7,1988. Prepared for ENEL, CRTN and EPRI; EPRI GS-6317, April, 1989. pp. 2.1-1 – 2.1-24

[59] LIFFICK, G. W. AND COOPER, J. W. (1995): Thermal Performance Upgrade of the Arkansas NuclearOne Cooling Tower: A Root Cause Analysis Approach. Paper presented at the 57th AnnualAmerican Power Conference. V. 57-2. pp. 1357-1362.

[60] LINDAHL, P. AND BUGLER, T. (1986): Counterflow Cooling Tower Design with High EnergyEvaluations. 5th IAHR Cooling Tower Workshop, Monterey, California, 1986.

[61] LUCAS, M. J. AND BUCHLIN, J. M. (1986): A Study of Cold Air Inflow into a Vertical Heated Tube.5th IAHR Cooling Tower Workshop, Doubletree Inn at Fisherman’s Wharf Hotel, Monterey,

REFERENCES

105

California, September 29-October 3, 1986. International Association for Hydraulic Research,Division III – Industrial and Structural Fluid Mechanics, Section on Fluid Mechanics in EnergyProduction. Proceedings prepared for the Electric Power Research Institute.

[62] MADADNIA, J., BERGSTROM, E. R. AND RENNIE, J. (1996): Wind Tunnel Studies of Methods toReduce Wind-Induced Cooling Tower Performance Losses. Presented in the Sixth InternationalConference in Heat and Mass Transfer.

[63] MADADNIA, J., REIZES, J., BEHNIA, M., REVEL, A. AND WEIDEMIER, R. (1998): PerformanceImproving Structures for Cooling Towers. Proceedings of the 11th IAHR Cooling TowerSymposium. Cottbus, Germany. 22 September, 1998.

[64] MADADNIA, J., REIZES, J., BEHNIA, M., COOMBES, P., KOOSHA, H., BOJNORDI, M., AL-WAKEED, R.AND HILL, W. (2001a): Wind Effects on the Operation of a Natural Drought Wet Cooling Tower; a)Performance Analysis. Proceedings of the 12th IAHR Symposium in Cooling Tower and HeatExchangers, Editors: Jafar Madadnia & Homa Koosha, University of Technology, Sydney, Facultyof Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.: M-101, pp. 8-21. ISBN: 1-86365-697-9.

[65] MADADNIA, J., BOJNORDI, M. AND KOOSHA, H. (2001b): Wind Effects on Performance of a NaturalDrought Wet Cooling Tower; b) Influence of the Packing Design (Scaled-model). Proceedings ofthe 12th IAHR Symposium in Cooling Tower and Heat Exchangers, Editors: Jafar Madadnia &Homa Koosha, University of Technology, Sydney, Faculty of Engineering, Sydney, Australia, 11-14November, 2001, Paper No.: W-102, pp. 228-234. ISBN: 1-86365-697-9.

[66] MADADNIA, J., BOJNORDI, M. AND KOOSHA, H. (2001c): Wind Effects on the Operation of a NaturalDrought Wet Cooling Tower; c) Effectiveness Analysis of Performance Improving Devices (Scaled-model). Proceedings of the 12th IAHR Symposium in Cooling Tower and Heat Exchangers, Editors:Jafar Madadnia & Homa Koosha, University of Technology, Sydney, Faculty of Engineering,Sydney, Australia, 11-14 November, 2001, Paper No.: S-100, pp. 326-333. ISBN: 1-86365-697-9.

[67] MADADNIA, J., KOOSHA, H., BOJNORDI, M. AND AL-WAKEED, R. (2001d): An Overview ofExperimental and Numerical Studies in Power Station Cooling Towers. Proceedings of the 12th

IAHR Symposium in Cooling Tower and Heat Exchangers, Editors: Jafar Madadnia & HomaKoosha, University of Technology, Sydney, Faculty of Engineering, Sydney, Australia, 11-14November, 2001, Paper No.: S-101, pp. 334-348. ISBN: 1-86365-697-9.

[68] MAJUMDAR, A. K., SINGHAL, A. K. AND BARTZ, J. A. (1984): Effect of the Plume on thePerformance of a Natural Draft Cooling Tower. AIChe Symposium Series, Heat Transfer – NiagaraFalls, No. 236, Vol. 80, 1984.

[69] MAJUMDAR, A. K., SINGHAL, A. K. AND SPALDING, D. B. (1983): Program for 2-D Analysis ofFlow, Heat and Mass Transfer in Evaporative Cooling Towers. Volume 1 and 2, CS-2923 Volume1: EPRI Research Project, March 1983.

[70] MANIER, G. AND BENESCH, W. (1977): Häufigkeitsverteilungen der Windgeschwindigkeit bis 250mHöhe für die Bundesrepublik Deutschland. Meteorol. Rundschau 30. S. 144-152.

[71] MANIER, G. (1975): Vergleich zwischen Ausbreitungsklassen und Temperaturgradienten. Meteorol.Rundschau 28. S. 6-11.

[72] MARKÓCZY, G. AND STÄMPFLI, E. (1977): Experimentelle Untersuchung des EinflussesAtmosphärischer Umgebungsbedingungen auf das Betriebsverhalten von Naturzug-Trockenkühltürmen. EIR-Bericht, No. 324, Würenlingen, Switzerland, 1977.

[73] MERKEL, F. (1926): Verdünstungskühlung. VDI Zeits. 70, (1), pp. 123-8.

[74] MESAROVIĆ, M. M. (1973): ARARAT – A Computer Code for Thermal Design of Cooling Towers.Nuclear Engineering and Design, Vol. 24, No. 1, January 1973.

REFERENCES

106

[75] MONJOIE, M., BOUTON, F. AND EROPE, H. T. (2000): A New Look at Recalculation AnalysisIncluding Measure on a 50 Cells Cooling Tower. Cooling Tower Institute, Paper No.: TP00-06, pp.1-25.

[76] MOORE, F. K. (1976): Dry Cooling Towers. Advances in Heat Transfer, Vol. 12. Academic Press,1976.

[77] MOORE, F. K. (1978): Aerodynamics of the Heat Exchangers and their Arrangement in Large DryCooling Towers. ASME Publ. 78-WA/HT-19, Winter Annual Meeting, Dec. 10-15, 1978.

[78] MOORE, F. K. AND GARDE, M. A. (1981): Aerodynamic Losses of Highly Flared Natural DraftCooling Towers. 3rd Waste Heat Management and Utilisation Conference, Miami Beach, May 11-13, 1981.

[79] MOUSSIOPOULOS, N. AND FLASSAK, T. (1989): A Diagnostic Wind Model and its Interactions withObservations and Experiments. In: Computers and Experiments in Fluid Flow (Carlomagno, G. M.and Brebbia, C. A., Eds.). Springer, Berlin, pp. 239-250.

[80] OKAMOTO, S. (1982): Turbulent Shear Flow behind Hemisphere-Cylinder Placed on Ground Plane.Turbulent Shear Flows, Vol. 3, Springer-Verlag, Berlin, Heidelberg, New York, pp. 171-185.

[81] OKAMOTO, S. AND SUNABASHIRI, Y. (1992): Vortex Shedding from a Circular Cylinder of FiniteLength Placed on a Ground Plane. Journal of Fluids Engineering, Vol. 114, pp. 512-521. December,1992.

[82] OKAMOTO, T. AND YAGITA, M. (1973): The Experimental Investigation on the Flow Past a CircularCylinder of Finite Length. Bulletin of Japan Society of Mechanical Engineers, Vol. 16, pp. 805-814.

[83] PÁLFALVI, G. (1989): Heller Cooling Towers in China. Conference on Cooling Systems withCooling Towers. Budapest, September 27-29, 1989. Published by the Scientific Society forBuilding, 89/1136 MTESZ Házinyomda, Budapest. pp. 27-42.

[84] PATANKAR, S. V. (1980): Numerical Heat Transfer and Fluid Flow. Hemisphere Publishing.Washington D. C. 1980.

[85] PEEARLMUTTER, D., ERELL, E., ETZION, Y., MEIR, I. A. AND DI., H. (1996): Refining the Use ofEvaporation in an Experimental Down-Draft Cool Tower. Energy and Building, Vol. 23, pp. 191-197.

[86] POPPE, M. (1972): Wärme- und Stoffübertragung bei der Verdünstungskühlung im Gegen- undKreuzstrom. Dissertation, Technische Universität Hannover, 1972.

[87] RADOSAVLJEVIĆ, D. AND SPALDING, D. B. (1989): Simultaneous Prediction of Internal and ExternalAerodynamic and Thermal Flow Fields of a Natural-Draught Cooling Tower in a Cross-Wind. In:Dinelli, G. and Bellagamba, B. (Eds.): Proceedings of the International Cooling-Tower Conferenceheld in Pisa, Italy, October 4-7, 1988. Prepared for ENEL, CRTN and EPRI; EPRI GS-6317, April,1989. pp. 1.1-1 – 1.1-20

[88] RAZAFINDRAKOTO, E., DENIS, C. AND BALESTRE, L.(1998): N3S-AERO: A MultidimensionalModel for Numerical Simulation of Flows in Cooling Towers. Proceedings of the 11th IAHRCooling Tower Symposium. Cottbus, Germany. 22 September, 1998.

[89] RAZAFINDRAKOTO, E. AND FOURNIER, Y. (1997): Projet N3S de Mécanique des fluides. INTN3S:Module d’interpolation de variables d’une grille à une autre. Note EDF/DER HE-41/97/010A, 1997.

[90] RAŽNJEVIĆ, K. (1964): Toplinske Tablice i Dijagrami. Tehnička Knjiga, Zagreb, 1964.

[91] REIZES, J. AND MADADNIA, J. (2001): Predictions of Prototype Cooling Tower Performance fromModel Studies. Proceedings of the 12th IAHR Symposium in Cooling Tower and Heat Exchangers,Editors: Jafar Madadnia & Homa Koosha, University of Technology, Sydney, Faculty of

REFERENCES

107

Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.: T-107, pp. 149-165. ISBN: 1-86365-697-9.

[92] RENNIE, E. J. AND HAY, N. (1990): Analysis of Full-Scale Natural-Draft Tower Measurements toDevelop and Validate Tower Performance Computer Codes. International Association for HydraulicResearch, 7th Cooling Tower and Spraying Pond Symposium, Leningrad, USSR, May 29-June 2,1990.

[93] RHIE, C. M. AND CHOW, W. L. (1983): Numerical Study of the Turbulence Flow past an Airfoil withTrailing Edge Separation. AIAA J. 21. pp. 1525-1532.

[94] RICHTER, E. (1969): Untersuchung der Strömungsvorgänge am Austritt von Naturzugkühltürmenund deren Einfluss auf die Kühlwirkung. dr. Ing. Thesis, TU Dresden, 1969.

[95] RISH, R. F. (1961): The Design of a Natural Draught Cooling Tower. Int. Dev. on Heat Transfer,Part V, pp. 954-958, A.S.M.E., New York, 1961.

[96] RODI, W. (1980): Turbulence Models for Environmental Problems. In: Prediction Methods forTurbulent Flows (Edited by Kollmann, W.), pp. 259-350. Hemisphere, New York.

[97] ROHSENOW, W. M. (1985a): Foundation Handbook of Heat Transfer Fundamentals, 2nd Edition.McGraw-Hill, New York, 1985.

[98] ROHSENOW, W. M. (1985b): Foundation Handbook of Heat Transfer Application, 2nd Edition.McGraw-Hill, New York, 1985.

[99] ROSTEN, H. I. AND SPALDING, D. B. (1987a): The PHOENICS Beginner’s Guide. CHAM TR/100,1987.

[100] ROSTEN, H. I. AND SPALDING, D. B. (1987b): The PHOENICS Reference Manual. CHAM TR/200,1987.

[101] ROSTEN, H. I. AND SPALDING, D. B. (1987c): The PHOENICS Equations. CHAM TR/99, 1987.

[102] RUSSELL, C. M. B., MCCHESNEY, H. R., HOLDER, D. W., JONES, T. V. AND VERLINDEN, M. (1978):Cross Wind and Internal Flow Characteristics of Dry Cooling Towers. Combustion, May, 1978.

[103] SABATON, M. (1984): Prediction of the Influence of Wind and Relief on the Operation of theCooling Towers of the Future CHOOZ B Power Station Using Measurements Recorded onIsothermal and Heating Scale-Models. 4th IAHR Cooling Tower Workshop, Switzerland, October 2-5, 1984.

[104] SAMALI, B. AND MADADNIA, J. (2001): Wind Simulation in an Environmental Wind Tunnel for bothStructure and Performance Studies. Proceedings of the 12th IAHR Symposium in Cooling Tower andHeat Exchangers, Editors: Jafar Madadnia & Homa Koosha, University of Technology, Sydney,Faculty of Engineering, Sydney, Australia, 11-14 November, 2001, Paper No.: M-105, pp. 44-52.ISBN: 1-86365-697-9.

[105] SCHLICHTING, H. (1965): Grenzschicht-Theorie. Verlag G. Braun, Karlsruhe, 1965.

[106] SLAOUTI, A. AND GERRARD, J. H. (1981): An Experimental Investigation of the End Effects on theWake of a Circular Cylinder Towed through Water at Low Reynolds Numbers. Journal of FluidMechanics, Vol. 112, pp. 297-314.

[107] SPALDING, D. B. (1981): A General Purpose Computer Program for Multi-Dimensional One andTwo Phase Flow. Mathematics and Computers in Simulation. North Holland Press, Vol. XXIII, pp.267-276, 1981.

[108] STONE, H. L. (1968): Iterative Solution of Implicit Approximation of Multidimensional PartialDifferential Equations. SIAM J. Numer. Anal. 5. pp. 530-558.

REFERENCES

108

[109] STRAW, M. (2001): Numerical Simulation of Atmospheric Processes. Oral presentation at theEducator’s Club of the Budapest University of Technology and Economics, Professional Lecturesand Consultation organised by the Department of Fluid Mechanics and the Committee onMeteorology of the Hungarian Academy of Sciences, 8th November, 2001.

[110] SU, M. D., TANG, G. F. AND FU, S. (1999): Numerical Simulation of Fluid Flow and ThermalPerformance of a Dry-Cooling Tower under Cross Wind Condition. Journal of Wind Engineeringand Industrial Aerodynamics 79. Elsevier Science Ltd. pp. 289-306.

[111] SZABÓ, Z. (2002): Private communication. EGI Contracting-Engineering Co. Ltd. 2002 December

[112] TA LUFT (1986): Erste Allgemeine Verwaltungsvorschrift zum Bundes-Immissionsschutzgesetz(Technische Anleitung zur Reinhaltung der Luft - TA Luft) vom 27.02.1986 (GMB1. S. 95, ber. S.202)

[113] VAN STADEN, M. P., HASSE, G. W. AND JANSE VAN RENSBURG, J. J. (1996): CFD Modelling of NaturalDraught Cooling Towers to Investigate the Effect of Wind on Tower Performance. AEROTHERMComputational Dynamics, Project No.: AT96-012. Prepared for GEA Air Cooled Systems, SouthAfrica and GEA EGI Budapest, Hungary. May 1996.

[114] VAUZANGES, M. AND RIBIER, G. (1984): Influence of the Heterogeneity of Air and Water Flow onthe Functioning of Cooling Towers. 4th IAHR Cooling Tower Workshop, Switzerland, 1984.

[115] VAUZANGES, M. AND RIBIER, G. (1986): Variation of the Head Losses in the Air Inlets of NaturalDraught Cooling Towers with the Shape of the Lintel and Shell Supports. 5th IAHR Cooling TowerWorkshop, Monterey, California, September 29-October 3, 1986. International Association forHydraulic Research, Division III – Industrial and Structural Fluid Mechanics, Section on FluidMechanics in Energy Production. Proceedings prepared for the Electric Power Research Institute.

[116] VDI 2047: Kühltürme; Begriffe und Definitionen.

[117] VDI 2049 (1981): Wärmetechnische Abnahme- und Leistungsversuche an Trockenkühltürmen.VDI-Gesellschaft Energietechnik, Ausschuß Trockenkühltürme. VDI-Verlag GmbH, Düsseldorf,1981.

[118] VDI 3782 BLATT 1 (2001): Umweltmeteorologie, Atmosphärische Ausbreitungsmodelle.Gauß’sches Fahnenmodell für Pläne zur Luftreinhaltung. VDI-Richtlinien, Verein DeutscherIngenieure, Düsseldorf, Dezember 2001. Available from Beuth Verlag GmbH, Berlin. ICS13.040.01; 13.040.20. In: TA Luft -Technische Anleitung zur Reinhaltung der Luft (VDI/DIN-Handbuch Reinhaltung der Luft, Band 1b), Ausgabe Dezember 2002.

[119] VDI 3784 BLATT 2 (1990): Umweltmeteorologie, Ausbreitungsrechnung bei Ableitung vonRauchgasen über Kühltürme. VDI-Richtlinien, Verein Deutscher Ingenieure, Düsseldorf, März1990. Available from Beuth Verlag GmbH, Berlin. 551.571.7::536.423.1 551.513.2 (083.132). VDI-Handbuch Reinhaltung der Luft, Band 1.

[120] VDI-WÄRMEATLAS (1977): Berechnungsblätter für den Wärmeübergang. 3. Aufl. Düsseldorf: VDI-Verl. 1977.

[121] VÖLLER, G. (1985): Untersuchungen zum Windeinfluß auf die Strömung im Naturzug-Trockenkühlturm, Doctor Thesis, Universität Hannover, 1985.

[122] WEI, Q., ZHANG, B., LIU, K., DU, X. AND MENG, X. (1995): A Study of the Unfavourable Effects ofWind on the Cooling Efficiency of Dry Cooling Towers. Journal of Wind Engineering and IndustrialAerodynamics 54/55. Elsevier Science B. V. pp. 633-643.

[123] WILLA, J. L., CAMPBELL, J. C. AND HOFFMANN, L. (1980): Common Misconceptions ConcerningCooling Tower Performance. Cooling Tower Institute, Paper No.: TP80-08, pp. 1-15.

REFERENCES

109

[124] WITTE, R. (1983): Das Betriebsverhalten Atmosphärisch Beeinflusster Trockenkühltürme. Fortschr.-Ber VDI-Z, Reihe 6, No. 132, VDI-Verlag, Düsseldorf, 1983.

[125] WOLF, C. (2001): Monitoring Cooling Tower Thermal Performance. Proceedings of the 12th IAHRSymposium in Cooling Tower and Heat Exchangers, Editors: Jafar Madadnia & Homa Koosha,University of Technology, Sydney, Faculty of Engineering, Sydney, Australia, 11-14 November,2001, Paper No.: M-106, pp. 53-59. ISBN: 1-86365-697-9.

[126] XU, Y., ZHAO, Z. AND LU, Z. (1989): Some Investigations of Cooling Towers in China. In: Dinelli,G. and Bellagamba, B. (Eds.): Proceedings of the International Cooling-Tower Conference held inPisa, Italy, October 4-7, 1988. Prepared for ENEL, CRTN and EPRI; EPRI GS-6317, April, 1989.pp. 1.8-1 – 1.8-14

[127] ZHU, J. (1991): A Low Diffusive and Oscillation-Free Convection Scheme. Commun. Appl. Numer.Meths. 7. pp. 225-232.

[128] ZHAO, Z., SHI, J. AND MAO, K. (1989): Experimental Study on the Aerodynamic Losses in NaturalDraught Cooling Towers. In: Dinelli, G. and Bellagamba, B. (Eds.): Proceedings of the InternationalCooling-Tower Conference held in Pisa, Italy, October 4-7, 1988. Prepared for ENEL, CRTN andEPRI; EPRI GS-6317, April, 1989. pp. 1.9-1 – 1.9-8

[129] ZHAO, Z. AND SHI, J. (1998): The Unfavorable Effect of Natural Wind on Natural Draft Dry CoolingTowers and its Improvement Measures. Proceedings of the 11th IAHR Cooling Tower Symposium.Cottbus, Germany. 22 September, 1998.

110

List of Own Publications23 February 2005

Journal papers in English, published in Hungary

1. Kapás, N., „Investigation of flow characteristics of Heller type cooling towers with differentcooling delta angles,” Periodica Polytechnica Series in Mechanical Engineering, Vol. 47, No. 2,pp. 143-150, 2003. Published by the Budapest University of Technology and Economics. HUISSN 0324-6051.

Presentations in English, published in proceedings of international conferences

2. Kapás, N., „Behaviour of natural draught cooling towers in wind,” Proceedings of the Conferenceon Modelling Fluid Flow – The 12th Event of International Conference Series on Fluid FlowTechnologies held in Budapest, Hungary, September 3-6, 2003. Edited by T. Lajos and J. Vad, pp.668-675. Published by Department of Fluid Mechanics, Budapest University of Technology andEconomics. ISBN 963 420 777 4ö.

Journal papers in Hungarian

3. Kapás, N., „A magyar távhőigények előrejelzése analóg módszerrel,” (Prediction of the Hungariandistrict heating demands by analogous technique), Magyar Energetika, Vol. 9, No. 4, pp. 19-24,August 2001. ISSN 1216-8599.

4. Kapás, N., „Energiahordozók és technológiák versenye a XXI. században,” (The competition ofenergy sources and technologies in the 21st century), Magyar Energetika, Vol. 10, No. 2, pp. 43-48, April 2002. ISSN 1216-8599.

5. Kapás, N., „Turbulens áramlások modellezése az atmoszférikus határrétegben,” (Turbulent flowmodelling in the atmospheric boundary layer), Magyar Energetika, Vol. 10, No. 6, pp. 31-35,December 2002. ISSN 1216-8599.

6. Kapás, N., „Kísérleti mérések a Heller-féle száraz hűtőtornyon,” (Experimental measurements onthe Heller-type dry cooling tower), Magyar Energetika, Vol. 11, No. 1, pp. 45-47, February 2003.ISSN 1216-8599.

7. Kristóf, G., Varga, L. and Kapás, N., „A numerikus áramlástan néhány energetikai alkalmazása,”(Some applications of numerical hydrodynamics in energetic), Magyar Energetika, Vol. 11, No. 2,pp. 23-26, April, 2003. ISSN 1216-8599.

Unpublished studies

Research reports

8. Csaba, G. and Kapás, N., „Heller rendszerű hűtőberendezések méréskiértékelő programja, a Bursa-i és az Al-Zara-i garanciamérések kiértékelése,” (Measurement evaluation program of Heller typecooling systems, evaluation of performance guarantee tests taken at Bursa and Al-Zara), internalreport, No. 6810-ECSR-T10. EGI - Contracting Engineering Co. Ltd. 7th August, 2002.

Oral presentations

9. Kapás, N., „Analyzing the energy supply in the 21st century,” at the Student Programme of the 18th

World Energy Congress & Exhibition, organised by the World Energy Council in Buenos Aires,Argentina, 21-25 October, 2001.

10. Kapás, N., „Application of CFD for thermal and flow modelling of Heller type cooling towers,” atthe series of professional programs organised jointly by the Department of Fluid Mechanics ofBUTE, Fluid Machinery Section of the Scientific Society of Machine Industry and the competentsubcommittee of the Hungarian Academy of Sciences. Theme: „Application of numericalsimulation of flows in energetic.” BUTE Department of Fluid Mechanics, December 10, 2002.

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

111

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

Figure App-1. Cup anemometer transmitter and Sher-disc

Figure App-2. Flow field measurement around the tower

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

112

Figure App-3. Cup anemometer and Sher-disc fixed onto a 5.5 m long rodbelayed beside the tower for data collection during night

Figure App-4. Cooling water flow rate measurement with the ultrasonic flow meter

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

113

Figure App-5. Siemens SDF water flow sonde

Figure App-6. Capacitive differential pressure transmitter (yellow) with power supply unitfor measuring the pressure difference of the water flow sonde

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

114

Figure App-7. PT 100 platinum resistance thermometers for measuring the temperature of cooling water

Figures App-8. and App-9. Mounting the PT 100 platinum resistance thermometers in front of theheat exchangers for dry bulb temperature measurement of ambient air entering the cooling tower

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

115

Figure App-10. Climbing the tower in order to install the thermometers on the top

Figure App-11. Mounting the PT 100 platinum resistance thermometers on the consolefor dry bulb temperature measurement of ambient air at the top of the tower

APPENDIX: PICTURES FROM THE ON-SITE MEASUREMENT

116

Figure App-12. The 1,400 MWe Bursa CCPP (Turkey)

Figure App-13. Bursa CCPP by night

117

ÖSSZEFOGLALÓ MAGYARUL

A szél hatása természetes huzatú hűtőtornyokra igen összetett jelenség. Ezek afolyamatok az atmoszférikus határrétegben játszódnak le, és hosszú ideig a szélhatásszámszerű értékének meghatározására a tornyokon végzett helyszíni mérések voltakegyedül megbízhatóak. Újabban igen gyors fejlődésnek lehettünk tanúi a numerikusáramlástan területén is, ami által bonyolult geometriák esetén is elvégezhetők azáramlástani számítások.

A doktori kutatómunkám célja új megoldásokat találni és elemezni az erőművitermészetes huzatú száraz hűtőtornyok szélben való viselkedésének javítására. Ha egyléghűtésű hűtőtorony szélnek van kitéve, az áramlási mező eltorzul a torony körül. Ennekkövetkeztében néhány hőcserélőn több hűtőlevegő áramlik keresztül, míg más hűtőkönkevesebb. A szél a hűtőtorony kilépő keresztmetszeténél is megzavarja az áramlást.Összességében véve a torony hatékonysága lecsökken, ami azt jelenti, hogy agőzkörfolyamat kondenzációs hőjét csak magasabb hűtővíz hőmérsékletek mellett lehet atoronyban leadni a környezetbe. Ez megnöveli a kondenzátorban a kondenzációshőmérsékletet és egyben a gőzturbina ellennyomását is, ami lerontja az erőmű termikushatásfokát és kapacitását – ugyanannyi villamos energia megtermeléséhez többtüzelőanyag szükséges, ugyanakkor a blokk villamos hálózat számára rendelkezésre állókapacitása lecsökken. A torony szél által lerontott hűtőképességéből származó költségekmeghatározhatók, melyek indokolttá tehetik olyan beavatkozások alkalmazását,melyekkel a szélhatás gazdaságosan csökkenthető.

Egy adott hűtőtorony aktuális szélhatása nagyszámú paramétertől függ, így a szélhatástnehéz általánosan meghatározni. Ezért a doktori kutatómunkám keretén belül egy konkréttermészetes huzatú erőművi hűtőtorony került kiválasztásra, melyen a vizsgálatokatelvégeztem. A hűtőtorony környezeti szélben való működésének tanulmányozásakoralkalmazott módszerek helyszíni mérésekből és korszerű numerikus áramlástaniszámításokból álltak. Az utóbbi módszer esetében a FLUENT nevű programothasználtam.

A doktori értekezésemben megtalálható a numerikus modell részletes leírása és azeredményül kapott áramlási mezők elemzése. A numerikus számításokkal meghatározottszélhatás görbét összevetettem a helyszíni mérésekből kapott görbével, ésmegállapítottam, hogy a két módszer igen jó egyezést mutatott.

Számos hűtőtorony szélhatást javító beavatkozást is bemutattam. A vizsgált esetekmagukba foglalták a növelt hűtővízáram, a hőcserélők közötti szabályozotthűtővízmennyiség elosztás, a szabályozott torony hűtőlégáram, terelőlemezek, valamint amódosított toronyhéj alak hatásainak elemzését. A műszaki vizsgálatokon túlgazdaságossági számításokat is végeztem a szélhatás javítása kapcsán a fenti esetekre.

A fő eredményeket öt tézisben foglaltam össze.

118

SUMMARY IN ENGLISH

Wind effect on natural draught cooling towers has complex physics. It takes place in theatmospheric boundary layer, and for a long time, the only reliable method to determine itsquantitative value has been the on-site measurement on full-scale towers. Recently, wehave seen a very rapid development also in the field of Computational Fluid Dynamics(CFD), whereby simulation of fluid flows in a complex geometry became feasible.

The purpose of my Ph.D. study was to find and analyse new solutions for improvingthe behaviour of natural draught power plant dry cooling towers in wind. When an air-cooled cooling tower is subjected to cross-wind, the flow field is distorted around thetower. As a consequence, some heat exchangers have higher cooling air flow throughthem, while other coolers have lower. The fluid flow is disturbed by the wind also at thecooling tower outlet opening. On the whole, the efficiency of the tower decreases, whichmeans that the condensation heat from the steam cycle can be rejected to the environmentonly with increased cooling water temperatures in the cooling tower. This results in ahigher temperature of condensation in the condenser and hereby a higher steam turbineback pressure, which cuts down the thermal efficiency and capacity of the power plant –the same output of electric energy will require more input of fuel, as well as the unit’scapacity, available for the electricity grid, decreases. The costs caused by the decreasedcooling plant capability in wind can be determined, and these costs could make reason-able the application of such accessories that can mitigate the wind effect economically.

Actual wind effects on specific cooling towers depend on a large number ofparameters, so the effect of wind is difficult to express in a general way. Therefore,within the frameworks of my Ph.D. work a specific cooling tower was selected to beanalysed. The methods included full-scale measurements in the field and state-of-the-artCFD investigations of the problems relating to the operation of the aforementionednatural draught power plant cooling tower in ambient winds. The applied ComputationalFluid Dynamics software was FLUENT.

In my Ph.D. dissertation the details of the computational model were described, theobtained flow fields were analysed, and the wind effect curve resulting from the CFDcalculations was compared with that obtained from full-scale site measurements. Theagreement between experimental and numerical results was more than satisfactory.

Several improving measures aimed at lowering the deteriorating effect of wind on thecooling tower thermal performance were also discussed. These improving measuresincluded increased cooling water flow, controlled cooling water flow among heatexchangers, controlled air flow through the tower, baffle plates and modified shape oftower shell. Beyond technical surveys, economical analysis of these measures was alsocarried out.

The main results are summarised in five theses.

119

ABSTRACT

The effects of wind on power plant natural draught dry cooling towers were investigatedby means of full-scale field measurements and Computational Fluid Dynamics calcula-tions. The relationship between ambient wind speed and the change in warm water tem-perature in cooling tower was established and several interventions aimed at improvingthe behaviour of the tower in wind were analysed. These improving measures includedincreased cooling water flow, controlled cooling water flow among heat exchangers,controlled air flow through the tower, baffle plates and modified shape of tower shell.Beyond technical surveys, economical analysis of these measures was also carried out.

A doktori értekezés címének fordítása angolról magyarra:

A szél hatásai a természetes huzatú száraz hűtőtornyokra